Accepted Manuscript

Title: Physiological and behavioral effects of animal-assisted

interventions on therapy dogs in pediatric oncology settings

Authors: Amy McCullough, Molly A. Jenkins, Ashleigh

Ruehrdanz, Mary Jo Gilmer, Janice Olson, Anjali Pawar,
Leslie Holley, Shirley Sierra-Rivera, Deborah E. Linder,
Danielle Pichette, Neil J. Grossman, Cynthia Hellman,
Noémie A. Guérin, Marguerite E. O’Haire

PII: S0168-1591(17)30330-1
DOI: https://doi.org/10.1016/j.applanim.2017.11.014
Reference: APPLAN 4560

To appear in: APPLAN

Received date: 19-7-2017

Revised date: 3-11-2017
Accepted date: 30-11-2017

Please cite this article as: McCullough, Amy, Jenkins, Molly A., Ruehrdanz, Ashleigh,
Gilmer, Mary Jo, Olson, Janice, Pawar, Anjali, Holley, Leslie, Sierra-Rivera, Shirley,
Linder, Deborah E., Pichette, Danielle, Grossman, Neil J., Hellman, Cynthia, Guérin,
Noémie A., O’Haire, Marguerite E., Physiological and behavioral effects of animal-
assisted interventions on therapy dogs in pediatric oncology settings.Applied Animal
Behaviour Science https://doi.org/10.1016/j.applanim.2017.11.014

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Physiological and behavioral effects of animal-assisted interventions on therapy dogs in pediatric
oncology settings

Authors: McCullough, Amy (PhD)a, Jenkins, Molly A. (MSW)a, Ruehrdanz, Ashleigh (MPH)a, Gilmer,
Mary Jo (PhD, RN-BC, FAAN)b, Olson, Janice (MD)c, Pawar, Anjali (MD)d, Holley, Leslie (MS, CCLS)e,
Sierra-Rivera, Shirley (BSN, RN)e, Linder, Deborah E. (DVM, MS, DACVN)f, Pichette, Danielle (CCRP)g,
Grossman, Neil J. (MD)g, Hellman, Cynthia (MSW)a, Guérin, Noémie A. (MS)h, & O’Haire, Marguerite E.
(PhD)h

PT
Affiliation postal addresses:
a
American Humane/1400 16th Street NW, Suite 360/Washington, DC 20036 USA.

RI
b
Monroe Carell Jr. Children's Hospital at Vanderbilt/417 Godchaux Hall/461 21st Avenue South/Nashville,
TN 37240 USA.

SC
c
Randall Children’s Hospital at Legacy Emanuel/2801 N. Gantenbein Avenue, 3rd Floor/Portland, OR
97227 USA.

U
d
UC Davis Comprehensive Cancer Center/4501 X Street, 2nd Floor/Sacramento, CA 95817 USA.
e

f
N
St. Joseph’s Children’s Hospital/3001 W. Dr. Martin Luther King Jr. Boulevard/Tampa, FL 33607 USA.
A
Cummings School of Veterinary Medicine at Tufts University/200 Westboro Road/North Grafton, MA
01536 USA.
M

g
UMass Memorial Children’s Medical Center/55 Lake Avenue North/Worcester, MA 01655 USA.
h
Center for the Human-Animal Bond/Purdue University, College of Veterinary Medicine/725 Harrison
D

Street/West Lafayette, IN 47907 USA.

TE

Author Email Addresses (in author order):

Amy McCullough (corresponding author): [email protected]; Molly A. Jenkins:

[email protected]; Ashleigh Ruehrdanz: [email protected]; Mary Jo
EP

Gilmer: [email protected]; Janice Olson: [email protected]; Anjali Pawar:

[email protected]; Leslie Holley: [email protected]; Shirley Sierra-Rivera:
[email protected]; Deborah E. Linder: [email protected]; Danielle Pichette:
[email protected]; Neil J. Grossman: [email protected];
CC

Cynthia Hellman: [email protected]; Noémie A. Guérin: [email protected]; and

Marguerite E. O’Haire: [email protected].

Corresponding author:
A

Amy McCullough, PhD

American Humane
1400 16th Street NW, Suite 360
Washington, DC 20036 USA

Phone: 303-588-6225
Fax: N/A
Email Address: [email protected]

1
Highlights:
 The responses of dogs to animal-assisted intervention sessions were investigated.
 Dogs’ salivary cortisol did not increase during animal-assisted intervention sessions.
 Dogs did not have increased stress-associated behaviors during AAI sessions.
 Dogs who exhibited more stress behaviors also exhibited more affiliative behaviors.
 Dogs who exhibited more stress behaviors also had increased salivary cortisol

PT
Abstract

Over the past two decades, animal-assisted interventions (AAIs), defined as the purposeful incorporation

RI
of specially trained animals in services to improve human health, have become increasingly popular in

SC
clinical settings. However, to date, there have been few rigorously-designed studies aimed at examining the

impact of AAIs on therapy animals, despite a notable potential for stress. The current study measured

U
physiological and behavioral stress indicators in therapy dogs who participated in AAI sessions in pediatric

N
oncology settings, while also examining the psychosocial effects for patients and their parents. This
A
manuscript describes the study’s canine stress findings. A total of 26 therapy dog-handler teams were paired

with newly diagnosed children with cancer at five children’s hospitals in the United States. These teams
M

provided regular AAI visits to the child and his/her parent(s) for a period of four months. The teams
D

completed a demographic form, the Canine Behavioral Assessment & Research Questionnaire (C-BARQ),
TE

and a self-report survey to document the types of activities that occurred during each session. Canine saliva

was also collected at five baseline time points and 20 minutes after the start of study sessions for cortisol
EP

analysis, and all study sessions were video recorded to document the dog’s behavior via an ethogram

measure. Data showed no significant differences in salivary cortisol levels between baseline (0.51µg/dL)
CC

and AAI sessions (0.44µg/dL; p = 0.757). Higher salivary cortisol was significantly associated with a higher

number of stress behaviors per session (p = 0.039). There was a significant relationship between stress and
A

affiliative session behaviors (p<0.0001), indicating that dogs who exhibited more stress behaviors also

exhibited more affiliative behaviors. The dog’s most commonly coded session behaviors were oral

behaviors, such as lip licking, and tail wagging. The only C-BARQ factor that was found to have a

significant association was stranger-directed fear; higher scores on this factor were significantly associated

2
with the dog exhibiting fewer affiliative behaviors in sessions (b = 2.12, p = 0.042). Results show that

therapy dogs did not have significantly increased physiological stress responses, nor did they exhibit

significantly more stress-related behaviors than affiliative-related behaviors, while participating in AAIs in

pediatric oncology settings. There was a significant relationship between canine cortisol and behavior, thus

strengthening the argument for the use of cortisol in canine well-being research. This study discusses the

PT
importance of further investigation to confirm these findings, and to enhance therapy dog involvement in

RI
hospital settings.

SC
Keywords: animal-assisted intervention, therapy dog, stress, salivary cortisol, behavior, animal welfare

U
1. Introduction
N
Documented benefits of human-animal interactions have increasingly led to the integration of animal-
A
assisted interventions (AAIs1) in services designed to improve human health, and to greater investment into
M

exploring their effects (McCardle et al., 2011). When included as adjuncts to medical and mental health

treatment, interactions with therapy animals have been shown to distract from or alleviate anxiety; decrease
D

pain perception; motivate therapeutic participation; normalize clinical environments and enhance therapist-
TE

client rapport; and provide social support (Braun et al., 2009; Fine, 2015; Nimer and Lundahl, 2007; Wu et

al., 2002). Dogs, because of their relative trainability, predictable behavior, and availability are commonly
EP

utilized as therapy animals in a variety of AAI applications and therapeutic settings (Glenk, 2017).

Through nearly 35,000 years of domestication, dogs have become well-attuned to nuances in human
CC

emotions, behaviors, and attentional states (Albuquerque et al., 2016; Reid, 2009; Wang et al., 2016).

Studies (Handlin et al., 2011; Odendaal and Meintjes, 2003) show that people who interact with dogs
A

experience elevated levels of the attachment hormone oxytocin, with recent research (Nagasawa et al.,

1
AAI is to be defined as the purposeful incorporation of specially trained and qualified animals, most commonly
dogs, in services to improve human health (International Association of Human-Animal Interaction Organizations or
IAHAIO, 2014; Nimer and Lundahl, 2007).

3
2015) demonstrating increased oxytocin for both humans and their pet dogs through mutual gazing. While

heightened sensitivity to the human condition is likely advantageous for connecting with AAI recipients, it

may also pose salient welfare risks to therapy dogs. For example, dogs have exhibited similar physiological

stress responses as humans after listening to infant crying (Yong and Ruffman, 2014). Because such

outward displays of human distress often occur in therapeutic settings, therapy dogs may be vulnerable to

PT
stress during AAI participation (Glenk, 2017). Additionally, as registered therapy dogs are trained to calmly

RI
tolerate interactions that they might otherwise find taxing (e.g., rushed head-petting by strangers), handlers

may find it difficult to notice distress in their dogs who may already be inclined to please (Glenk, 2017).

SC
To date, research regarding the experience of therapy dogs in AAIs has been informative, but remains

limited. Most measures of therapy dog welfare rely on physiological (cortisol), behavioral (stress-related)

U
and/or observational (handler perspective) indicators, as it is problematic to assess canine stress by any one
N
indicator alone. Without behavioral context, it is difficult to conclude whether elevated cortisol indicates
A
distress or eustress (Edgar et al., 2012).
M

While current research suggests minimal welfare concerns for therapy dogs, results have been mixed and

are difficult to generalize given a lack of practice fidelity. Likewise, organizations that register therapy dogs
D

often differ in their health and safety policies, including inconsistent vetting and training procedures meant
TE

to limit canine stress (Linder et al., 2017). Recent research (Haubenhofer and Kirchengast, 2007, 2006)

found that therapy dogs experience higher levels of physiological stress on working days when compared
EP

to days at home. Salivary cortisol was also higher for dogs participating in shorter and more frequent AAI

sessions, possibly due to heightened intensity or fewer respite opportunities (Glenk, 2017; Haubenhofer
CC

and Kirchengast, 2007, 2006). King et al. (2011) reported indicators of canine physiological and behavioral

stress after 1-2 hours of AAI participation in hospital settings, including stress-related behaviors, increased
A

salivary cortisol, and cortisol-stress behavior correlation. Additionally, a higher prevalence of behavioral

stress has been observed in therapy dogs during interactions with children younger than 12 years, whose

relatively erratic behavior may cause dogs discomfort (Marinelli et al., 2009).

4
Environmental factors (e.g., strange stimuli/people) may also increase therapy dog stress. Recently,

researchers found that novel settings significantly increased salivary cortisol in therapy dogs when

compared to familiar locations, underlining the importance of allowing dogs to become accustomed to new

AAI environments (Ng et al., 2014). Likewise, handlers have reported such factors as high temperatures

and confined spaces as stressful for dogs (Marinelli et al., 2009).

PT
In contrast, other studies have found scant evidence indicative of canine distress during AAIs (Barstad,

RI
2014; Palestrini et al., 2017; Piva et al., 2008), including no differences in physiological stress between

working and non-working days (Glenk et al., 2014, 2013; Ng et al., 2014). In research examining cortisol

SC
and behavior, Glenk et al. (2014) found that therapy dog behavior did not significantly change over time,

and that salivary cortisol decreased in the final two sessions with adults undergoing substance abuse

U
treatment.
N
Simply put, AAI practice must be mutually beneficial to be considered both ethical and effective, and
A
handlers must continually work to ensure the welfare needs and humane treatment of participating therapy
M

animals (Serpell et al., 2010). Given the potential for canine stress during AAIs, as well as the current lack

of agreement regarding if and how AAIs affect animal welfare, additional research is essential to understand
D

how to best support therapy dogs as their involvement in human health services broadens. To that end, this
TE

study measured the physiological and behavioral effects of regular AAI sessions for registered therapy dogs

in five U.S. pediatric oncology settings. This research also measured the effects of sessions on patients and
EP

their parents, with these findings reported elsewhere (McCullough et al., in press). In regard to canine

outcomes, we hypothesized that therapy dogs would exhibit minimal indicators of physiological and
CC

behavioral distress over their study participation.

2. Methods
A

All protocols and surveys were approved by the Institutional Animal Care and Use Committee and the

Institutional Review Board at American Humane, as well as at each participating hospital site. The

following five hospitals, located in the United States, participated in this research study: Monroe Carell Jr.

Children’s Hospital at Vanderbilt in Nashville, Tennessee; Randall Children’s Hospital in Portland,

5
Oregon; University of California Davis Children’s Hospital in Sacramento, California; St. Joseph’s

Children’s Hospital in Tampa, Florida; and Children's Medical Center at UMass Memorial Health Care, in

partnership with Cummings School of Veterinary Medicine at Tufts University, in Worcester,

Massachusetts. Modifications to the protocol were made at two different time points, which impacted the

human participants’ enrollment and retention rates (e.g., broader inclusion criteria for children and a

PT
monetary incentive for control group patients). These changes were not related to any of the canine

RI
protocols.

2.1 Participants

SC
Eligible therapy dog-handler teams were identified by the hospital sites’ volunteer services coordinator,

study coordinator, and/or by other handlers through a referral process. Each participating team consisted of

U
a therapy dog and his/her handler, who were registered by a national organization that requires rigorous
N
training and evaluation policies. These teams volunteered to visit children, ages 3-17 years, who had been
A
recently diagnosed with cancer, enrolled in the study at one of the study sites, and randomly assigned to the
M

intervention group to receive regular (e.g., weekly) visits from a therapy dog-handler team for a period of

four months. Prior to participation, all handlers were required to complete the hospital volunteer services
D

training and credentialing process, and to adhere to the rigorous health, safety, and privacy guidelines and
TE

policies in place at their respective hospital site. To support safe interactions, all therapy dog handlers had

completed extensive AAI training, and all therapy dogs had passed a behavioral and health evaluation, prior
EP

to enrollment. Dogs were required to be a minimum of one year old and have at least six months of AAI

experience with the handler prior to enrolling in the study.

CC

Twenty-six teams participated in the study. Prior to consenting, each team attended a study-specific training

at their local hospital site. At this meeting, a detailed description of the study protocol, including how to
A

collect their dog’s saliva and how to complete all other applicable instruments, was presented. Handlers

had the opportunity at this training to practice and demonstrate their ability to properly obtain a saliva

sample from their dog. All handlers received printed training materials to keep for their reference, as well

as online links to videos demonstrating proper saliva collection techniques.

6
Sample size was determined based on a power analysis for the human measures in the study, which yielded

a sample size of 100 patients to be enrolled. Approximately half of this group (n = 60) was randomly

assigned to receive AAI sessions with therapy dogs. To facilitate these sessions for 60 child participants

over a period of 33 months, 26 dogs participated in the canine assessment portion of the study.

2.2 Animal-Assisted Intervention

PT
. Human participants randomized to this study’s intervention group received their standard care offered at

RI
the hospital plus regular visits from a participating therapy dog-handler team. Children were enrolled for a

four-month period near the onset of their cancer treatment. Interactions with the therapy dog occurred

SC
approximately once per week, and coincided with the child’s treatment schedule and clinic visits.

Depending on existing hospital policies and procedures, children transferred to inpatient care could

U
continue their visits with the therapy dog, while others could visit in the outpatient oncology clinic only.
N
Children were matched, by the site coordinator, to a specific therapy dog-handler team who became their
A
primary team for the study; back-up teams were used on an as-needed basis. Children were matched to their
M

therapy dog team based on the child’s treatment schedule and the team’s volunteer availability.

All therapy dog sessions were prescribed to last approximately 20 minutes, with session length ultimately
D

determined by the patient, parent, handler, site coordinator, or another medical staff member. AAI sessions
TE

were nonprescriptive in nature, allowing for flexibility in activities and physical intensity based upon the

child’s health status and the level of therapy dog and child engagement. However, all handlers were given
EP

a list of commonly used activities to choose from as a reference guide (e.g., pet the dog, brushed the dog,

practiced dog’s cues, etc.). These sessions took place in semi-private to private areas of the pediatric
CC

oncology clinic, depending upon the health status of the child and room availability.

2.3 Measures
A

A variety of instruments were used to assess indicators of canine stress and well-being over the course of

the study.

2.3.1 Demographic Form

7
Demographic information was obtained from participating handlers using a self-report questionnaire. In

addition to factors such as age, race, gender, and canine breed, handlers reported information about the

length of time they had been volunteering with their therapy dog, with what populations, and in what types

of settings.

2.3.2 Canine Behavioral Assessment & Research Questionnaire (C-BARQ)

PT
After consenting to participate, therapy dog handlers completed the C-BARQ for their dog who would be

RI
participating in the study. This standardized 100-question instrument is completed online and asks dog

owners to indicate how their dog typically responds to common events and stimuli in their environment

SC
(Serpell, 2017). The C-BARQ is comprised of 14 behavioral factors: trainability, stranger-directed

aggression, owner-directed aggression, dog rivalry, stranger-directed fear, non-social fear, dog-directed

U
aggression, dog-directed fear, touch sensitivity, separation-related behavior, excitability,
N
attachment/attention-seeking, chasing, and energy. This instrument was utilized as a descriptor for the
A
therapy dog population’s temperament and behavior.
M

2.3.3 Salivary Cortisol

Salivary cortisol is a known bio-marker for stress and arousal in humans and other animals (Hekman et al.,
D

2012; Kirschbaum and Hellhammer, 1994). Dog’s salivary cortisol levels lag plasma levels by 20 minutes,
TE

indicating that salivary cortisol peaks roughly 20 minutes after a stressful event, which is then maintained

for approximately 0.5 hours before declining (Hennessy et al., 1998; Vincent and Michell, 1992). In this
EP

study, therapy dog-handler teams collected their dog’s saliva at five different time points at the beginning

of the study to establish a baseline cortisol measurement. These five time points were as follows: 1. Upon
CC

their dog waking in the morning, prior to breakfast; 2. Mid-day/noon; 3. Evening, just prior to the dog’s

typical bedtime; 4. Approximately 20 minutes after the presence of an AAI-specific “trigger” – such as the
A

introduction of his/her therapy vest/bandanna or visit bag; and 5. Approximately 20 minutes after arriving

at the participating hospital/clinic site.

Given the known salivary cortisol lag and peak times, as well as our flexible approach to session timing

described above, the study protocol specified that therapy dog handlers collect their canine’s saliva sample

8
immediately following a session, except for in cases where sessions lasted fewer than 20 minutes. In

instances where sessions lasted less than 20 minutes, handlers were asked to wait in a quiet room for a short

period to allow for the saliva they collected to accurately represent their dog’s response to the session (e.g.,

if a session lasted five minutes, the handler waited in a quiet room for a period of 15 minutes and then began

the saliva collection process).

PT
Saliva was collected by the therapy dog handler using recommended collection procedures (Dreschel and

RI
Granger, 2009). To collect saliva, the handler sat/kneeled on the floor or a chair, and placed an absorbent

125mm long Salimetrics (State College, PA, USA) SalivaBio Children’s Swab into the dog’s mouth. The

SC
handler would rub the swab around the dog’s cheek and gum area for approximately 90-300 seconds, until

the swab appeared to be saturated with saliva. Saturated swabs were then placed into a saliva collection

U
tube pre-labeled with the Handler ID, Child ID (for post-session samples), and sample number. Handlers
N
then hand-wrote the date and time of collection, and gave the saliva collection tube to the study coordinator
A
to store and ship for analysis. All tubes were stored at -20° C in a medical grade freezer at each hospital to
M

ensure sample viability. Approximately once every three months, a set of samples were shipped to

Salimetrics, in a Styrofoam cooler packed with dry ice, for testing. Each sample underwent
D

Salimetrics Cortisol Enzyme Immunoassay Kit, which is a duplicate assay. The results of the assays were
TE

averaged to create one value for the study analyses, unless the saturation amount would not allow for

duplicate, and were calculated as µg/dL.

EP

2.3.4 Therapy Dog Handler and Study Coordinator Self-Reports

After each AAI session, handlers completed a self-report form to indicate the activities that occurred. They
CC

were provided with a list of common AAI activities to choose from, but could also indicate if any other

activities took place that were not pre-defined. This list of session activities was tracked for both the child
A

and parent using one-zero sampling (i.e., activity occurred or did not occur). Handlers also used this form

to describe any environmental factors related to the session and their dogs’ behavior (e.g., “room was very

warm which affected dog's attention and behavior”, “[dog] intermittently distracted by noise in hallway”,

etc.). The site coordinator filled in who was present and participated in the activities, and indicated any

9
observations of the child’s or dog’s behavior that they felt was pertinent (e.g., “had to meet with dog in

hallway today”, “child quite nauseated today”, etc.).

2.3.5 Canine Behavior Ethogram

Each AAI session was recorded by two video cameras (Samsung HMX-F90BN HD Camcorder with 2.7"

LCD Screen) to ensure that the dog’s behavior was captured during the interaction. These recordings were

PT
then coded using a non-standardized behavior ethogram developed by the study’s researchers and informed

RI
by animal behaviorists and veterinarians. This ethogram included 26 behaviors divided into three

categories: affiliative indicators (leaning or resting body or head against a person or object; licking a

SC
person; pawing/paw lifting; play stance/bow; pushing snout; raising ears; rolling over; tail wagging;

walking towards a person), moderate stress indicators (body shaking or "shaking off"; escape; looking

U
at/to handler; looking away; oral behaviors, such as lip licking, extending the tongue, or smacking the lips;
N
panting - excessive or prolonged; restlessness; self-directed behaviors; yawning), and high-stress
A
indicators (baring teeth; barking, yelping, yipping, whining, or whimpering; biting or attempting to bite;
M

crouching; drooling--excessive or in copious amounts; ears plastered/pinned back; growling; stare gaze)

(See Table 2 for detailed ethogram information). All behaviors exhibited by the therapy dogs were tallied
D

for their frequency. The total frequency of stress-related behaviors was summed to create one score (17
TE

items), as were the affiliative-related behaviors (9 items).

Ethogram coders underwent several rounds of training on how to use the instrument, including viewing
EP

video clips of the potential behaviors that were to be coded. Intra-class correlation coefficients (ICC)were

calculated at various time points during the training process, and additional training sessions were
CC

conducted until a satisfactory level of inter-rater agreement and coder confidence was established. One

individual served as the primary coder, with two additional coders providing supplementary coding support.
A

2.4 Settings

Data collection for this study occurred in the handler’s place of residence and at the participating hospital

sites.

2.4.1 Home Setting

10
Four out of the five baseline cortisol data points (morning, mid-day, evening, trigger) were obtained at the

therapy dog team’s place of primary residence. The morning collection occurred upon the dog waking, and

typically occurred between 05:19 and 10:30 (M = 07:47, SD = 01:13 hours:minutes). The mid-day

collection occurred between 07:00 and 21:15 (M = 12:40, SD = 2:41 hours:minutes). The evening collection

occurred between 10:40 and 23:00 (M = 19:57, SD = 3:16 hours:minutes). The trigger collection occurred

PT
approximately 20 minutes after giving the dog a signal that s/he would be going to work (e.g., therapy dog

RI
vest/bandanna). The trigger sample was collected between 07:05 and 23:00. (M = 11:11, SD = 3:12

hours:minutes).

SC
2.4.2 Hospital Setting

The final baseline saliva sample was collected 20 minutes after arriving at the hospital where the team

U
participated in study sessions. This collection occurred between 08:15 and 16:45 (M = 12:16 SD = 2:08
N
hours:minutes). The goal of collecting baseline samples at the dog’s home and at the hospital was to obtain
A
a measure of the dog’s typical salivary cortisol levels on non-working days (without any interactions with
M

unfamiliar individuals) to then compare to cortisol collected after each AAI session.

To allow for the therapy dog to have minimal distractions during the collection process, post-session saliva
D

samples were collected at the hospital site in a quiet, semi-private to private area designated by the site
TE

coordinator and handler. This area was generally separate from where the AAI sessions took place.

2.5 Statistical Analyses

EP

Analyses were conducted to address two key goals: 1). to compare differences in salivary cortisol between

control/baseline and AAI conditions and 2). to examine factors affecting salivary cortisol and behavior
CC

during AAI sessions. Statistical analyses were conducted using mixed modeling procedures (Raudenbush

and Bryk, 2002). This hierarchical analysis technique accounted for the nested data design, with repeated
A

measurements nested within individual dogs nested within hospital sites. Random factors in all models

included dog and hospital site.

Salivary cortisol data were positively skewed and not normally distributed. Therefore, raw values were log-

transformed for further analyses using linear mixed models. Means and graphs present back-transformed

11
values for ease of interpretation. All models for salivary cortisol control for time of day as a fixed factor.

To compare the effect of study conditions on salivary cortisol, we conducted a linear mixed model with the

fixed effect of condition (morning, mid-day, evening, trigger, hospital, AAI session). To examine factors

affecting salivary cortisol, we conducted linear mixed models with additional fixed factors related to

demographics (age, gender, experience, breed), C-BARQ factor scores, or AAI session characteristics

PT
(duration, number of people present and participating, reported activities).

RI
Ethogram behavior data were analyzed using generalized linear mixed models appropriate for count data

with a Poisson distribution sampling method and a log-link function. To control for the differing video

SC
lengths, all ethogram behavior models controlled for the length of the video as a fixed factor. To examine

factors affecting behavior, we conducted linear mixed models for stress and affiliative behaviors with fixed

U
factors related to demographics (age, gender, experience, breed), C-BARQ factor scores, or AAI session
N
characteristics (number of people, reported activities). We also included the fixed factor of cortisol to
A
examine the relationship between cortisol and behavior.
M

Analyses were performed using the Statistical Package for the Social Sciences (SPSS) Version 24.0 (IBM

Corp., released 2016. IBM SPSS Statistics for Windows, Version 24.0. IBM Corp., Armonk, NY). All
D

significance tests were two-tailed with a significance level of α < 0.05.

TE

3. Results

The final dataset for analysis included 604 data points, or records, across 26 dogs. At the dog level (n = 26),
EP

data included demographic characteristics (n = 24 dogs) and C-BARQ surveys (n = 26 dogs). At the

repeated measures record level (n = 604), data included AAI Therapy Dog Handler and Study Coordinator
CC

Self-Reports (n = 445), ethogram behavioral data (n = 405), and valid salivary cortisol samples (n = 411).

The random factors in the mixed models accounted for variance across dogs and hospital sites. The random
A

factor of hospital site was not significant in any model (ICC ≤ 0.423, p ≥ 0.209), indicating that there were

no significant differences in outcomes across hospitals. The random factor of dog was significant in all

models (ICC ≥ 0.030, p ≤ 0.046), indicating that there were significant individual differences across dogs

that were accounted for in the analyses.

12
3.1 Participants

The sample of therapy dog-handler teams with demographic data (n = 24; two handlers declined to provide

this information) included dogs aged two to 13 years old, who were primarily female (58%) with 0.5 to

9.17 years of experience as a visiting therapy dog. A wide range of dog breeds of varying sizes was

represented, including Miniature Poodles, Newfoundlands, a Border Collie mix, a Dachshund, and a

PT
Wheaton Terrier, while the most common breeds were Golden Retrievers (17%), Labrador Retrievers

RI
(13%), or mixed breeds that included part Golden or Labrador Retriever (16%). Therapy dog handlers

identified as predominantly female (92%), 46 years or older (83%), and Caucasian/White (96%). Therapy

SC
dog-handler teams were largely registered through Pet Partners (75%). Dog handlers had between 0.5 and

10.42 years of experience as AAI handlers, with most of these teams (67%) having previous experience in

U
hospital settings.
N
3.2 Session Characteristics: Therapy Dog Handler and Study Coordinator Self-Reports
A
The study included 445 AAI sessions recorded by Therapy Dog Handler and Site Coordinator Self-Reports.
M

Sessions lasted between five and 180 minutes, with an average of 23.95 minutes per session (Median = 17;

Mode = 15; SD = 21.13). The categories of individuals at each session were noted and the number of people
D

who were present and participated at each session (excluding the handler) ranged from one to eight. On
TE

average, 3.11 people were present and 2.27 people participated in session activities with the therapy dog.

The most common people present included the child (n = 445 sessions, 100%), mother (n = 377, 85%),
EP

father (n = 151, 34%), and nurse (n = 110, 25%). The most common people who participated in session

activities included the child (n = 417, 94%), mother (n = 306, 69%), and father (n = 113, 25%).
CC

The most common activities engaged in by the child included petting the dog (n = 409, 92%), talking to the

dog (n = 307, 69%), viewing the dog’s photos (n = 144, 32%), and brushing the dog (n = 106, 24%). The
A

most common activities engaged in by the parent included petting the dog (n = 244, 55%), talking to the

dog (n = 212, 48%), taking a photo of the dog and/or child with the dog (n = 69, 16%), and viewing the

dog’s photos (n = 64, 14%). Across 445 recorded sessions, children participated in 1,446 activities, while

parents participated in 801 activities. See Table 1 for a complete list of session activity frequencies.

13
3.3 Canine Behavior Ethogram

AAI sessions were video recorded to code canine behavior via a pre-defined ethogram. Inter-rater reliability

of ethogram behavior data was calculated using a two-way mixed effect ICC. Reliability between the

primary and secondary coder was good (ICC = 0.782, p < 0.0001). More than 30 unique ethogram samples

were used to calculate the ICC, and therefore it is suggested that ICC “values between 0.75 and 0.9 indicate

PT
good reliability” (Koo and Li, 2016, p. 158). An average of 9.69 (SD = 10.47) stress behaviors and 9.14

RI
(SD = 9.33) affiliative behaviors were recorded in each session. There were no significant differences

between the frequency of displayed stress versus affiliative behaviors, t(404) = 1.19, p = 0.236. The most

SC
common canine behaviors included oral behaviors, such as lip licking, and tail wagging. There were no

high-stress indicator behaviors coded for any dog across all study sessions. Descriptive statistics for

U
ethogram behaviors are presented in Table 3. There was a significant association between stress and
N
affiliative behaviors, β = 0.21, t(402) = 9.39, p < 0.0001. Dogs who displayed more stress behaviors in a
A
given session also demonstrated more affiliative behaviors during that session.
M

3.4 Salivary Cortisol

Salivary cortisol levels were relatively consistent across all conditions (Figure 1). There were no significant
D

differences between the AAI sessions (n=299) and any baseline session, including morning (n = 24, β = -
TE

0.05, p = 0.398), mid-day (n = 23, β = 0.02, p = 0.680), evening (n = 22, β = 0.08, p = 0.295), trigger (n =

25, β = -0.04, p = 0.507), and hospital (n = 22, β = 0.03, p = 0.637)

EP

Canine cortisol levels did not increase over time of involvement in the study, and remained consistent with

their baseline levels.

CC

3.5 Relationship Between Demographics, Salivary Cortisol, and Behavior

The relationship between demographic characteristics and canine outcomes are presented in Table 4.
A

Therapy dog gender was related to salivary cortisol (p = 0.002), with female dogs exhibiting significantly

lower salivary cortisol than males. Animal age was related to salivary cortisol (p = 0.004), stress behaviors

(p = <0.0001), and affiliative behaviors (p = 0.010). Older dogs showed lower salivary cortisol and

displayed more behaviors during AAI sessions, including both stress and affiliative behaviors. There were

14
no effects of handler years of experience on salivary cortisol or behavioral outcomes (p range = 0.197 -

0.421).

There was no relationship between most C-BARQ factors and canine outcomes (p range = 0.129 - 0.882).

The only factor with a significant effect was stranger-directed fear. Higher stranger-directed fear, as rated

by handlers, was associated with 2.12 fewer affiliative behaviors on average per session (p = 0.042).

PT
3.6 Relationship Between Session Characteristics, Salivary Cortisol, and Behavior

RI
The relationship between session characteristics and canine outcomes is presented in Table 5. Duration of

sessions was not significantly related to salivary cortisol (p = 0.481) or stress behaviors (p = 0.092). Session

SC
length was significantly associated with affiliative behaviors (p < 0.0001), but the effect was relatively

small; therapy dogs displayed 0.01 more affiliative behaviors for each minute longer a session lasted. There

U
were no significant effects of the number of people present (p ≥ 0.165) or participating (p ≥ 0.134) on
N
salivary cortisol or behavior. There were also no effects of the type of session activities on salivary cortisol
A
(p ≥ 0.105).
M

There were significant relationships between session activity type and behavior (Table 6). Dogs displayed

significantly fewer affiliative behaviors per session when the following activities took place: participant got
D

water for the dog; child played a game on the dog’s vest; child drew a picture of the dog; child used a
TE

stethoscope to listen to the dog’s heartbeat; child collected an AAI-related sticker or card from the handler;

participant used a lint roller during the session; participant brushed the dog; and participants discussed
EP

therapy dogs with the handler (p range = ≤ 0.001 - 0.043).

Dogs displayed significantly more affiliative behaviors per session when the following activities took place:
CC

participant played with the dog’s toy; participant talked to the dog; handler discussed dog breeds and

histories with the family; participants took the dog for a walk; participants viewed photos of the dog; and
A

participants practiced the dog’s known cues (p range = ≤ 0.001 - 0.046).

Dogs exhibited significantly fewer stress behaviors per session when the following activities took place:

child facilitated a visit between the therapy dog and hospital staff; participants pet the dog; and participants

took a photo of the dog (p range = ≤ 0.001 - 0.041). Finally, dogs displayed significantly more stress

15
behaviors per session when the following activities took place: child put a bandanna on the dog and when

the handler and participants discussed dog breeds (p = 0.035 and p = 0.003, respectively).

3.7 Relationship Between Salivary Cortisol and Behavior

The final analysis model included salivary cortisol level as a fixed factor to examine its relationship with

stress and affiliative behaviors. When controlling for affiliative behaviors, higher salivary cortisol was

PT
significantly associated with an increased frequency of stress behaviors per session, β = 0.16, t(243) = 2.08,

RI
p = 0.039. When controlling for stress behaviors, higher salivary cortisol was significantly associated with

a reduced frequency of affiliative behaviors, β = -0.26, t(243) = -2.82, p = 0.005. Thus, lower levels of

SC
salivary cortisol were related to increased displays of affiliative behaviors during AAI sessions, while

higher levels of salivary cortisol were related to increased displays of stress behaviors.

U
4. Discussion
N
The current study measured physiological and behavioral indicators of stress in registered therapy dogs
A
who participated in AAI sessions with pediatric oncology patients and their parents. To our knowledge, this
M

is the largest randomized controlled trial to measure the impact of AAI for both humans and therapy dogs

to date, particularly across multiple hospitals where AAI is commonly practiced. As hypothesized, results
D

indicate that therapy dogs show minimal signs of distress during AAI sessions, regardless of hospital site.
TE

The geographical heterogeneity (across the U.S.) of participating sites has favorable implications for the

generalizability of these findings, particularly in healthcare facilities with rigorous therapy dog
EP

programming (Linder et al., 2017).

The lack of significant differences in canine salivary cortisol levels between AAI sessions and each of the
CC

five baseline conditions is an important finding that supports existing selected studies in this area (Glenk et

al., 2014, 2013; Ng et al., 2014). Cortisol concentration levels also remained relatively consistent over time,
A

which could indicate that participation in a greater number of AAI sessions is not a source of stress for

therapy dogs.

Additionally, there was a significant relationship between canine cortisol and behavior, with lower cortisol

levels associated with increased displays of affiliative behavior, and higher cortisol levels with increased

16
stress behaviors. Both of these relationships, but particularly the latter, suggest that cortisol may be a good

indicator of distress in therapy dogs. Notably, there were few significant relationships between handler-

rated canine temperament and canine outcomes, except for the C-BARQ stranger-directed fear factor; dogs

with a perceived higher degree of stranger-directed fear exhibited significantly fewer affiliative behaviors

on average per session. Much like with novel settings (Ng et al., 2014), therapy dogs who find interactions

PT
with strangers stressful may benefit from a prolonged introduction to unfamiliar recipients, possibly

RI
comprised of shorter initial visits where the dog can initiate contact and take breaks when needed. Further,

canine temperament measures like the C-BARQ may serve as effective tools for facilities when screening

SC
prospective therapy dog-handler teams, and determining their optimal placement.

In addition, while therapy dogs exhibited slightly more stress behaviors than affiliative behaviors per

U
session on average, the difference between the two was not significant and may have been due to more
N
stress behaviors listed on the ethogram for researchers to code. Dogs who displayed a high frequency of
A
affiliative behaviors tended to exhibit a similarly high frequency of stress behaviors, which could indicate
M

greater expressiveness in these dogs.

Interestingly, older dogs displayed significantly more stress and affiliative behaviors than their younger
D

counterparts, but also had significantly lower session cortisol. Potentially, some dogs may grow
TE

increasingly expressive in their behavior or adept at communicating behaviorally as they age. Further,

behaviors traditionally deemed as stress indicators – such as panting or yawning (Beerda et al., 1999, 1998)
EP

– could serve as a calming mechanism for dogs to manage their stress during worrisome situations, as Glenk

et al. (2014) propose. Thus, this could lead to a higher number of stress behaviors, but a lower physiological
CC

stress response, as we observed with older therapy dog participants. With that said, it is essential that

handlers be responsive to the changing health needs of their therapy dogs as they age, and retire dogs when
A

they are no longer able to engage comfortably with recipients (Serpell et al., 2010).

Years of handler experience in AAI was not significantly related to either their dog’s physiological or

behavioral outcomes; just as many canine stress indicators were present with more experienced handlers as

with those who were newer to the practice. For less seasoned handlers, fewer years of experience could

17
have been offset by a shorter time lapse since undergoing training. As AAI trainings often emphasize the

primary role of handlers as stewards of their dog’s safety and comfort (Pet Partners, 2017), new handlers

may have been more keenly attentive to canine behavior than those more accustomed to the practice. This

finding highlights the importance of regular handler training opportunities that emphasize how to

effectively monitor and alleviate canine stress during sessions. Nevertheless, it can be argued that handler

PT
participants overall were heedful of their dog’s needs, as evidenced by the significant relationship between

RI
longer sessions and an increased frequency of affiliative canine behaviors. Handlers who observed a

relatively high frequency of affiliative behaviors in their dogs may have continued or even prolonged AAI

SC
sessions due to the perceived low risk of canine stress. However, given the small effect size of this

relationship, and the large range of documented session duration, this interpretation remains speculative

U
and needs further exploration.
N
There were several limitations of this study that may serve to clarify areas of future inquiry. While our
A
canine sample size is the largest known to date in studies of therapy dog welfare (Glenk, 2017), we studied
M

less than 30 dogs. Subsequent research would benefit from a larger sampling of therapy dog participants.

However, the sample was strengthened by its heterogeneity and representation of the breed variability
D

currently observed in AAI applications.

TE

There were also certain outside factors that could have impacted the level of stress experienced by therapy

dogs that we did not document, including their adoption experiences, trauma histories, or any record of
EP

serious health conditions or medication regimens (although all dogs received prior veterinary approval for

AAI participation). We also did not note characteristics of AAI visits that the dogs may have participated
CC

in between study sessions, such as length, number, frequency, and population served. Indeed, due to ethical

reasons, therapy dogs were not prohibited from visiting with other people that they encountered when on
A

hospital premises for study sessions, and we did not inquire about these interactions.

Additionally, handler participants, while anecdotally representative, were not demographically diverse

enough to explore the potential effects of handler race, gender, or age on therapy dog well-being. Future

research and practice efforts should focus on the potential impact of handler demographics on canine and

18
human recipient outcomes, as well as how this population may be diversified to ensure culturally competent

AAI practices. Furthermore, handler adherence to certain study protocols, such as the timing of baseline

saliva collection, was not consistent, and could have impacted the interpretation of those data.

Further, in order to study the canine effects of AAI sessions as they are currently practiced, this study did

not prescribe a structured protocol of activities. Rather, the flow and type of session activities occurred at

PT
the discretion of patients, parents, handlers, and/or medical staff, and were largely dependent on child health

RI
status and/or the response of the child or dog to the interaction. Activities were documented, including how

frequently they were engaged by both children and their parents.

SC
While there were no significant effects of session activities on canine salivary cortisol, there were

significant relationships between certain activities and the frequency of affiliative and stress behaviors. For

U
example, sessions where the therapy dog was brushed may have been more stressful for the dog, as
N
evidenced by the presence of less affiliative behaviors. Likewise, dogs displayed more affiliative behaviors,
A
and potentially less stress, in sessions where they were taken for a walk. However, at present, it is not sound
M

to infer a direct causal relationship between any particular activity and the dogs’ response. Additionally,

the direction of the activity and associated canine behavior is not clear (i.e., did the dog show more
D

affiliative behaviors when walked or was the dog walked because he/she showed more affiliative
TE

behaviors?). These research questions should be explored further, as they may help to refine prescribed

protocols for future AAI research and practice.

EP

Finally, there may have been potential issues regarding the behavior ethogram used in this study. Because

a validated instrument for assessing therapy dog behavior during AAI sessions does not currently exist,
CC

piloting and utilizing our own measure was necessary. In developing our ethogram, it was important to refer

to previous canine behavioral measures (i.e., Beerda et al., 1998), consult with animal behaviorists and
A

veterinarians, and provide further context through the inclusion of both stress and affiliative behaviors.

Nevertheless, the development of a validated measure which re-evaluates the nature of currently recognized

behavioral indicators of therapy dog stress and affiliation in an AAI context is important to confirm this

study’s findings and strengthen those of future, related inquiries.

19
4.1 Conclusions

The increasing prevalence of therapy dog programs in today’s pediatric hospitals (Chubak and Hawkes,

2016) and other services merits further investigation regarding the well-being of dogs who work in these

settings. AAIs have traditionally been positioned as services to improve human health, but the welfare needs

of therapy animals must be elevated if their participation in this pursuit is to advance. While further research

PT
is needed to verify these results, this study provides valuable and rigorous evidence that AAI participation

RI
in pediatric hospitals does not appear to place registered therapy dogs at significant risk of stress.

SC
Funding: This work was supported by Zoetis; Morris Animal Foundation exclusively from a partnership

with the Human-Animal Bond Research Institute (D14HA-012, 2014); Newman’s Own Foundation; and

U
the Nora Roberts Foundation. The above funding sources did not have a role in the study design; in the
N
collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the
A
article for publication.
M

Acknowledgments: The authors wish to thank all the children, families, therapy dogs, handlers, hospitals,

and partners who participated in this study and made this research possible.
D

Conflicts of interest: none.

TE
EP
CC
A

20
References

Albuquerque, N., Guo, K., Wilkinson, A., Savalli, C., Otta, E., Mills, D., 2016. Dogs recognize dog and
human emotions. Biol. Lett., 12, 1-5, https://doi.org/10.1098/rsbl.2015.0883.

Barstad, B.N., unpublished results. Evaluation of animal welfare in dogs working with animal assisted
interventions for elderly people with dementia. 2014 Master thesis, Norwegian University of Life Sciences,
Ås, Norway. http://azs.no/wp-content/uploads/2016/09/Barstad-2014_Evaluation-of-animal-welfare-in-
dogs-working-with-animal-assisted-interventions-for-elderly-people-with-dementia.pdf (accessed 09

PT
March 2017).

Beerda, B., Schilder, M.B.H., van Hooff, J.A.R.A.M., de Vries, H.W., Mol, J.A., 1998. Behavioural,

RI
saliva cortisol and heart rate responses to different types of stimuli in dogs. Appl. Anim. Behav. Sci., 58,
365-381, https://doi.org/10.1016/S0168-1591(97)00145-7.

SC
Beerda, B., Schilder, M.B.H., van Hooff, J.A.R.A.M., de Vries, H.W., Mol, J.A., 1999. Chronic stress in
dogs subjected to social and spatial restriction. 1. Behavioral responses. Physiol. Behav., 66, 233-242,
https://doi.org/10.1016/S0031-9384(98)00289-3.

U
Braun, C., Stangler, T., Narveson, J., Pettingell, S., 2009. Animal-assisted therapy as a pain relief
intervention for children. Complement. Ther. Clin. Pract., 15, 105-109,
https://doi.org/10.1016/j.ctcp.2009.02.008.
N
A
Chubak, J., Hawkes, R., 2016. Animal-assisted activities: Results from a survey of top-ranked pediatric
oncology hospitals. J. Pediatr. Oncol. Nurs., 33, 289-296, https://doi.org/10.1177/1043454215614961.
M

Dreschel, N.A., Granger, D.A., 2009. Methods of collection for salivary cortisol measurement in dogs.
Horm. Behav., 55, 163–168, https://doi.org/10.1016/j.yhbeh.2008.09.010.
D

Edgar, J.L., Nicol, C.J., Clark, C.C.A., Paul, E.S., 2012. Measuring empathic responses in animals. Appl.
Anim. Behav. Sci., 138, 182-193, https://doi.org/10.1016/j.applanim.2012.02.006.
TE

Fine, A.H. (Ed.), 2015. Handbook on animal-assisted therapy: Foundations and guidelines for animal-
assisted interventions, fourth ed. Elsevier-Academic Press, San Diego, CA.
EP

Glenk, L.M., 2017. Current perspectives on therapy dog welfare in animal-assisted interventions. Animals,
7, 1-17, https://doi.org/10.3390/ani7020007.

Glenk, L.M., Kothgassner, O.D., Stetina, B.U., Palme, R., Kepplinger, B., Baran, H., 2014. Salivary cortisol
CC

and behavior in therapy dogs during animal-assisted interventions: A pilot study. J. Vet. Behav., 9, 98-106,
https://doi.org/10.1016/j.jveb.2014.02.005.

Glenk, L.M., Kothgassner, O.D., Stetina, B.U., Palme, R., Kepplinger, B., Baran, H., 2013. Therapy dogs'
A

salivary cortisol levels vary during animal-assisted interventions. Anim. Welf., 22, 369-378,
https://doi.org/10.7120/09627286.22.3.369.

Handlin, L., Hydbring-Sandberg, E., Nilsson, A., Ejdebäck, M., Jansson, A., Uvnäs-Moberg, K., 2011.
Short-term interaction between dogs and their owners: Effects on oxytocin, cortisol, insulin and heart rate—
an exploratory study. Anthrozoös, 24, 301-315, https://doi.org/10.2752/175303711X13045914865385.

Haubenhofer, D.K., Kirchengast, S., 2007. Dog handlers’ and dogs’ emotional and cortisol secretion

21
responses associated with animal-assisted therapy sessions. Soc. Anim., 15, 127–150,
https://doi.org/10.1163/156853007X187090.

Haubenhofer, D.K., Kirchengast, S., 2006. Physiological arousal for companion dogs working with their
owners in animal-assisted activities and animal-assisted therapy. J. Appl. Anim. Welf. Sci., 9, 165-172,
https://doi.org/10.1207/s15327604jaws0902_5.

Hekman, J.P., Karas, A.Z., Dreschel, N.A., 2012. Salivary cortisol concentrations and behavior in a
population of healthy dogs hospitalized for elective procedures. Appl. Anim. Behav. Sci., 141, 149–157,

PT
https://doi.org/10.1016/j.applanim.2012.08.007.

Hennessy, M.B., Williams, M.T., Miller, D.D., Douglas, C.W., Voith, V.L., 1998. Influence of male and

RI
female petters on plasma cortisol and behaviour: Can human interaction reduce the stress of dogs in a public
animal shelter? Appl. Anim. Behav. Sci., 61, 63-77, https://doi.org/10.1016/S0168-1591(98)00179-8.

SC
IAHAIO, 2014. White paper: The IAHAIO definitions for animal assisted intervention and guidelines for
wellness of animals involved. http://iahaio.org/best-practice/white-paper-on-animal-assisted-interventions/
(accessed 27 October 2017).

U
King, C., Watters, J., Mungre, S., 2011. Effect of a time-out session with working animal-assisted therapy
dogs. J. Vet. Behav., 6, 232–238, https://doi.org/10.1016/j.jveb.2011.01.007.
N
Kirschbaum, C., Hellhammer, D.H., 1994. Salivary cortisol in psychoneuroendocrine research: Recent
A
developments and applications. Psychoneuroendocrinology, 19, 313–333, https://doi.org/10.1016/0306-
4530(94)90013-2.
M

Koo, T.K., Li, M.Y., 2016. A guideline of selecting and reporting intraclass correlation coefficients for
reliability research. J. Chiropr. Med., 15, 155-163, https://doi.org/10.1016/j.jcm.2016.02.012.
D

Linder, D.E., Siebens, H.C., Mueller, M.K., Gibbs, D.M., Freeman, L.M., 2017. Animal-assisted
interventions: A national survey of health and safety policies in hospitals, eldercare facilities, and therapy
TE

animal organizations. Am. J. Infect. Control, 45, 883-887, https://doi.org/10.1016/j.ajic.2017.04.287.

Marinelli, L., Normando, S., Siliprandi, C., Salvadoretti, M., Mongillo, P., 2009. Dog assisted interventions
in a specialized centre and potential concerns for animal welfare. Vet. Res. Commun., 33, S93-S95,
EP

https://doi.org/10.1007/s11259-009-9256-x.

McCardle, P., McCune, S., Griffin, J.A., Esposito, L., Freund, L.S. (Eds.), 2011. Animals in our lives:
Human-animal interaction in family, community, and therapeutic settings, first ed. Paul H. Brookes
CC

Publishing Co, Baltimore, MD.

McCullough, A., Ruehrdanz, A., Jenkins, M.A., Gilmer, M.J., Olson, J., Pawar, A., Holley, L., Sierra-
Rivera, S., Linder, D.E., Pichette, D., Grossman, N.J., Hellman, C., Guérin, N.A., O’Haire, M.E., in press.
A

Measuring the effects of an animal-assisted intervention for pediatric oncology patients and their parents:
A multi-site randomized controlled trial. J. Pediatr. Oncol. Nurs.

Nagasawa, M., Mitsui, S., En, S., Ohtani, N., Ohta, M., Sakuma, Y., Onaka, T., Mogi, K., Kikusui, T,
2015. Oxytocin-gaze positive loop and the coevolution of human-dog bonds. Science, 348, 333-336,
https://doi.org/10.1126/science.1261022.

22
Ng, Z.Y., Pierce, B.J., Otto, C.M., Buechner-Maxwell, V.A., Siracusa, C., Werre, S.R., 2014. The effect of
dog-human interaction on cortisol and behavior in registered animal-assisted activity dogs. Appl. Anim.
Behav. Sci., 159, 69-81, https://doi.org/10.1016/j.applanim.2014.07.009.

Nimer, J., Lundahl, B., 2007. Animal-assisted therapy: A meta-analysis. Anthrozoös, 20, 225-238,
https://doi.org/10.2752/089279307X224773.

Odendaal, J.S.J., Meintjes, R.A., 2003. Neurophysiological correlates of affiliative behaviour between
humans and dogs. Vet. J., 165, 296–301, https://doi.org/10.1016/S1090-0233(02)00237-X.

PT
Palestrini, C., Calcaterra, V., Cannas, S., Talamonti, Z., Papotti, F., Buttram, D., Pelizzo, G., 2017. Stress
level evaluation in a dog during animal-assisted therapy in pediatric surgery. J. Vet. Behav, 17, 44–49,

RI
https://doi.org/10.1016/j.jveb.2016.09.003.

Pet Partners, 2017. Program requirements. https://petpartners.org/volunteer/become-a-handler/program-

SC
requirements/ (accessed 04 April 2017).

Piva, E., Liverani, V., Accorsi, P.A., Sarli, G., Gandini, G., 2008. Welfare in a shelter dog rehomed with
Alzheimer patients. J. Vet. Behav., 3, 87–94, https://doi.org/10.1016/j.jveb.2007.08.004.

U
Raudenbush, S.W., Bryk, A.S., 2002. Hierarchical linear models: Applications and data analysis methods,
second ed. Sage Publications, Thousand Oaks, CA.
N
Reid, P.J., 2009. Adapting to the human world: Dogs’ responsiveness to our social cues. Behav. Process.,
A
80, 325-333, https://doi.org/10.1016/j.beproc.2008.11.002.
M

Serpell, J.A., 2017. Canine behavioral assessment & research questionnaire (C-BARQ).
https://vetapps.vet.upenn.edu/cbarq/ (accessed 17 January 2017).
D

Serpell, J.A., Coppinger, R., Fine, A.H., Peralta, J.M., 2010. Welfare considerations in therapy and
assistance animals, in: Fine, A.H. (Ed.), Handbook on animal-assisted therapy: Theoretical foundations and
TE

guidelines for practice, third ed. Elsevier-Academic Press, San Diego, CA, pp. 481-503,
https://doi.org/10.1016/B978-0-12-381453-1.10023-6

Vincent, I.C., Michell, A.R., 1992. Comparison of cortisol concentrations in saliva and plasma of dogs.
EP

Res. Vet. Sci., 53, 342-345, https://doi.org/10.1016/0034-5288(92)90137-Q.

Wang, G.D., Zhai, W., Yang, H.C., Wang, L., Zhong, L., Liu, Y.H., Fan, R.X., Yin, T.T., Zhu, C.L.,
Poyarkov, A.D., Irwin, D.M., Hytönen, M.K., Lohi, H., Wu, C.I., Savolainen, P., Zhang, Y.P., 2016. Out
CC

of southern East Asia: The natural history of domestic dogs across the world. Cell Res., 26, 21-33,
https://doi.org/10.1038/cr.2015.147.

Wu, A.S., Niedra, R., Pendergast, L., McCrindle, B.W., 2002. Acceptability and impact of pet visitation
A

on a pediatric cardiology inpatient unit. J. Pediatr. Nurs., 17, 354-362,

https://doi.org/10.1053/jpdn.2002.127173.

Yong, M.H., Ruffman, T., 2014. Emotional contagion: Dogs and humans show a similar physiological
response to human infant crying. Behav. Process., 108, 155-165,
https://doi.org/10.1016/j.beproc.2014.10.006.

23
Figure 1 Back-transformed salivary cortisol concentrations across collection times

0.400

0.350
Mean Cortisol (µg/dL)

0.300

PT
0.250

0.200

RI
0.150

SC
0.100

0.050

U
0.000
N
Hospital
Evening

Session
Mid-Day

Trigger
Morning

A
M

Baseline AAI
D
TE
EP
CC
A

24
Table 1 Number, percentage, and types of activities that children and parents participated in during animal-assisted intervention
(AAI) sessions (n=445)

Activities Child Parent

n (number of sessions % (percentage of n (number of sessions % (percentage of
in which the activity sessions in which the in which the activity sessions in which the
occurred) activity occurred) occurred) activity occurred)
Pet the Dog 409 92 244 55
Talk to the

PT
Dog 307 69 212 48
View Dog’s
Photos 144 32 64 14
Brush the

RI
Dog 106 24 16 4
Practice the
Dog’s Cues 99 22 29 7

SC
Discuss Dog
Breeds 62 14 55 12
Discuss
Therapy

U
Dogs 56 13 48 11
Walk the
Dog 53 12 10 2
Take Photo
of the Dog 43 10
N 69 16
A
Play with
Dog’s Toy 42 9 15 3
M
Facilitate
Dog’s Visit
with Staff 24 5 16 4
Collect a
D

Sticker from
Handler 22 5 2 0
TE

Use Lint
Roller 18 4 4 1
Listen to
Dog’s Heart
EP

Beat 13 3 3 1
Teach Dog
New Trick 10 2 5 1
Color a
CC

Picture 8 2 0 0
Put Bandana
on the Dog 8 2 3 1
Get the Dog
Water 6 1 0 0
A

Draw a
Picture of or
for the Dog 6 1 0 0
Read to the
Dog 4 1 3 1
Play a Game 3 1 1 0
Play on
Dog’s Vest 3 1 2 0

25
Table 2 Canine Behavior Ethogram

BEHAVIOR Description How to Score

Affiliative Indicators
Leaning or resting body or Leaning or resting body or head against Count each time the dog leans or puts
head against a person or a person or object head down. If dog moves away, then
object replaces head or leans again, count
again. Dog is seeking contact, not just
resting.

PT
Licking a person Passing the tongue over any part of the Count each period of licking, i.e. if
person's body multiple licks over and over, just count
once. Then if dog stops licking and
starts again, count again.

RI
Pawing/paw lifting Forearm lifted to a 45 degree angle; Count each time paw is lifted. If dog
paw extended or “waved”; sometimes puts paw down and then lifts again,
paw touches the person count again. Dog is seeking attention.

SC
Play stance/bow Excited and alert affect; bottom raised; Count each time the dog bows.
tail raised and wagging; front of body
lowered; front knees bent; tongue out;
head forward and erect; ears perked up;
eyes wide and bright; dog may jump

U
around and vocalize in anticipation
Pushing snout/Seeking pet Pushing, investigating, or eliciting This is typically a dog nosing a person's
N
contact with the snout at any body part
of the person; "goosing," shoving, or
hand for petting. Count once for each
time the dog "asks" for more petting.
A
poking a person (usually the handler or
someone familiar to the dog)
Raising ears (breed-specific) Ears noticeably raise or perk up This could be in response to seeing a
M

person, hearing an interesting noise, or

responding to the person calling the dog
or squeaking a toy. The dog is showing
D

friendly interest. Count once for each

time the dog perks his ears.
Rolling over Rolling over on back, exposing Although this behavior can be either
TE

abdomen; may be accompanied with stress related or affiliative, typically a

body or limb stretching and/or self- therapy dog will roll over to seek
directed behaviors (e.g., scratching) petting. Count once for each time the
dog rolls over.
EP

Tail wagging Tail moves repeatedly side to side or up Count once for each period of tail
and down wagging. If tail stops wagging but then
resumes, count again.
Walking/Approach Walking forward; walking towards a Count once for each time the dog
CC

person approaches the client. This will likely

be just upon entrance.
Moderate Stress Indicators
Body shaking or "shaking Body shaking or trembling May happen after being hugged or
A

off" involuntarily OR "shaking off" close contact. Count once for each
voluntarily (like when wet or dirty) shaking episode.
Escape Efforts to get away from the eliciting May signal the dog is done visiting -
stimulus by pulling on leash, backing tries to get off the bed and/or walks
up, hiding behind handler, digging, etc. away from client (e.g., away from
client, toward the door). Count once
each time dog initiates the motion.

1
Looking at/to handler Frequent looking at/to handler during Dog seems unsure of what s/he is
interaction with client/patient; gazing at supposed to do - looks to the handler for
handler during interaction guidance or direction. Count once for
each time the dog looks up/at the
handler.
Looking away Head turning away from the person; Dog purposefully turns away from
averting eyes/gaze from the person client who may be trying to get close to
dog's face. Count once each time the
dog turns his/her head.
Oral Behaviors/Lip Licking Tongue out; tongue briefly extended; Count once when you see dog's tongue

PT
lip licking; snout/nose licking; floor extend from his/her mouth. Count as a
licking; swallowing; lip smacking licking period - not each individual
lick. For each period of 5 licks or more,
count as one period of licking.

RI
Panting - excessive or Breathing quickly or in a labored Count once for each panting period. If
prolonged fashion; tongue usually out; abdomen dog stops and resumes, count again.
may noticeably move up and down

SC
Restlessness Frequent changes in posture or Seems dog needs to readjust position -
position; frequent “changes in the state may be physically uncomfortable.
of locomotion”; circling; difficulty Could be coupled with trying to leave
sitting or lying still the bed/room and move toward the

U
door. Count once each time the motion
occurs.
Self-directed behaviors
chewing N
Grooming; scratching; licking; biting; Dog interrupts visiting with client to
chew/scratch on self. Count once for
each chewing/licking/scratching
A
episode.
Yawning Open mouth; inhalation of breath/air Count once each time you see the dog
M

followed quickly by exhalation yawn unless it's a repetitive period.

High Stress Indicators

Baring teeth Pulling the upper lips up and back so Typically paired with growling,
D

the teeth are visible (lip "curling"); looking as if dog will bite.
snarling
TE

Barking, yelping, yipping, Relatively brief vocalizations of Not 'talking' on command or vocalizing
whining, or whimpering varying pitch without growly for attention - aggressive or
undertones stressed/unhappy vocalizations.
Biting or attempting to bite Mouth open; teeth exposed; head Different than mouthing during play -
EP

forward; clamping (or attempting to aggressive action against patient or

clamp) the skin of a person between the others.
jaws; may cause noticeable wound on
person who is bitten
CC

Crouching Body crouched low; legs bent; bottom Dog is/looks backed into a corner and
and head lowered; back arched; tail does not want a person to approach.
may be between hind legs; cowering
Drooling--excessive or in Increased salivation or moisture around Not food-induced - unusual or
A

copious amounts the nose and mouth excessive drooling.

Ears plastered/pinned back Ears positioned lower and/or backward Sometimes paired with a submissive
(in response to stimulus) approach and may be followed by
rolling over.
Growling A throaty and rumbling vocalization, Dog makes a 'warning' sound that s/he
usually low in pitch feels protective/stressed.

2
Stare gaze An intense, fixed, and direct gaze into Part of a prey instinct -dog looks as if
the eyes of the person; eyes may be s/he is hunting.
dilated or the whites of the eyes may be
clearly visible

PT
RI
SC
U
N
A
M
D
TE
EP
CC
A

3
Table 3 Frequency of recorded animal behaviors during animal-assisted intervention (AAI) sessions (n=405)

Per Session Per Minute

Standard Deviation (# of times Standard Deviation (# of times
Behavior
Mean coded per session) Mean coded per session)
Stress 9.69 10.47 0.61 0.59
Affiliative 9.14 9.33 0.59 0.56
Oral Behaviors 4.52 7.28 0.28 0.42

PT
Licking 2.31 3.46 0.15 0.21
Tail Wagging 1.98 3.25 0.13 0.23
Yawning 1.39 1.84 0.09 0.11

RI
Raising Ears 1.24 3.89 0.07 0.21
Panting 1.17 1.36 0.08 0.10

SC
Snout (Pushing snout,
seeking pet) 1.09 1.67 0.07 0.12
Leaning (On the client) 0.97 1.71 0.06 0.10
Lifting Paw 0.97 2.13 0.06 0.14

U
Looking at Handler 0.79 1.32 0.05 0.09
Shaking (Shaking off
body, like when wet)
Walking
0.79
0.46
1.21
0.81
N 0.05
0.03
0.08
0.06
A
Restlessness (Frequent
posture changes) 0.29 1.25 0.02 0.07
M

Looking Away
(Intentionally looking
away from client) 0.26 1.16 0.02 0.07
Self-Directed Behaviors
D

(Grooming, chewing,
etc.) 0.25 0.78 0.02 0.05
TE

Escape (Attempting to
get away from
client/session) 0.21 0.82 0.01 0.04
Rolling Over 0.09 0.42 0.01 0.04
EP

Playing: Bow Stance 0.03 0.19 0.00 0.01

Vocalizing 0.01 0.11 0.00 0.01
Biting 0.00 0.00 0.00 0.00
CC

Crouching 0.00 0.00 0.00 0.00

Drooling (Excessively) 0.00 0.00 0.00 0.00
Ears Pinned Back 0.00 0.05 0.00 0.00
A

Growling 0.00 0.00 0.00 0.00

Staring (Fixed gaze at
client’s eyes) 0.00 0.00 0.00 0.00
Teeth (Baring Teeth) 0.00 0.05 0.00 0.00

4
Table 4 Relationship between demographic characteristics and canine salivary cortisol and behaviors (N = 406 samples, 23 dogs)

Cortisol Stress Behaviors Affiliative Behaviors

b SE b SE b SE
Demographic Characteristic

Dog Gender (Female vs. Male) -0.28** 0.09 -0.16 0.17 0.05 0.25
Dog Age (Years) -0.05* 0.02 0.15** 0.03 0.14** 0.04
Handler Experience (Years) -0.01 0.01 -0.04 0.03 0.06 0.04

PT
Breed (Lab/Golden vs. Other) 0.25* 0.09 -0.22 0.17 0.70** 0.25
Note: *p ≤ 0.05, **p ≤ 0.01, b = beta, regression coefficient, SE = standard error.

RI
SC
U
N
A
M
D
TE
EP
CC
A

5
Table 5 Relationship between session characteristics and salivary cortisol (N = 254 samples, 23 dogs) and ethogram stress and
affiliative behaviors (N = 368 samples, 25 dogs)

Stress Affiliative
Cortisol
Behaviors Behaviors
Session Characteristic B SE B SE B SE
0.0 0.0 0.0
Length of Session 0.00 0 0.00 0 0.01** 0
0.0 0.0 0.0

PT
Number of people present -0.03 2 -0.02 2 0.04 2
Number of people 0.0 0.0 0.0
participating 0.03 2 -0.01 2 0.03 2
Note: *p ≤ 0.05, **p ≤ 0.01, b = beta, regression coefficient, SE = standard error.

RI
SC
U
N
A
M
D
TE
EP
CC
A

6
Table 6 Relationship between session activities and coded ethogram stress and affiliative behaviors (N =368 samples, 25 dogs)

+ Affiliative Behaviors + Stress Behaviors

Played with dog’s toy (b = 0.47; p < .0001. Put a bandanna on the dog (b = 0.26; p = .035)

Talked to the dog (b = 0.26; p < .0001) Discussed dog breeds and histories (b = 0.15; p =

Discussed dog breeds and histories (b= 0.16; p = .003).

PT
.002)

Took the dog for a walk (b = 0.14; p = .046)

RI
Viewed photos of the dog (b= 0.13; p = .006).

SC
Practiced the dog’s known cues (b = 0.11; p= .033).

- Affiliative Behaviors - Stress Behaviors

U
Brushed the dog (b = 0.15; p = .003) Facilitated visit between dog and staff (b = 0.45; p

Got water for the dog (b = 1.32; p = 0.006) N

< .0001)
A
Used a lint roller (b = .20; p = .030) Pet the dog (b = 0.17; p = .034)
M
Drew a picture of the dog (b = 0.45; p = 0.003) Took a photo of the dog (b = 0.10; p = .041).

Listened to the dog’s heart beat (b = 0.33; p =

D

0.004)
TE

Collected a therapy dog sticker (b = 0.23; p =

0.043)
EP

Played dog board or card game (b = 1.00; p <

0.001)
CC

Discussed therapy dogs (b = 0.14; p = 0.020)

Note: b=beta, regression coefficient

A