Animal-Assisted Interventions for Autism Spectrum Disorder: A Systematic Review of the Literature from 2016 to 2020

Abstract

In the last five years, the literature on animal-assisted intervention (AAI) for Autism Spectrum Disorder (ASD) has doubled in size from 42 studies prior to 2015 to 85 studies (cumulative total assessed in 2020). Horses remain the most commonly researched animal for AAI, followed by dogs. The most commonly researched outcome was social interaction, across 21 studies. Though the quantity of studies has increased, issues with methodological rigor remain. Results highlight the need for a continued focus on methodological rigor as well as refining the structure of animal-assisted interventions, addressing the welfare needs of the animals involved, and continuing to establish an evidence base of both significant and nonsignificant findings for AAI for individuals with ASD.

Individuals with Autism Spectrum Disorder (ASD) are neurodiverse and can present with relative strengths in analytic-thinking, objective evaluation, systems design, and large knowledge bases within areas of special interest. A diagnosis of ASD, can include relative strengths (like the ones noted above) and relative difficulties in social communication and language learning (American Psychiatric Association, 2013). Additionally, individuals with ASD also present with intense interests and/or repetitive behaviors. Interventions designed to support individuals with ASD, include behavioral, educational, diet and lifestyle, pharmacologic treatments, and a multitude of complementary and integrative health options. Early behavioral interventions have the largest research base with robust empirical support for gains in language, social communication, and daily living skills (Polittle et al., 2015). However, these early behavioral models are not universally efficacious and families often seek additional/complementary treatment options. With 28% to 95% of families utilizing complementary treatment options, careful consideration of their efficacy and empirical support are warranted (Hofer et al., 2017). Within this review we focus on one commonly used and recently expanding complementary treatment option - animal-assisted interventions (AAI).

Studies of AAI, interventions that incorporate an animal, document both positive psychological and physiological effects with multiple theories and processes implicated (Beetz, 2017). For example, the Biophilia Hypothesis (Wilson, 1984) suggests that humans are innately drawn to nature and animals as a means of survival. The “biophilia effect” (Julius et al., 2013) suggests that when animals and nature are calm, individuals interpret safety in the environment. Animals may also offer social support to individuals serving as a nonjudgmental and secure attachment partner (Beetz, 2017). Animals can function as a bridge for social interactions in the human-to-human context offering a unique support for individuals with ASD (McNicholas & Collis, 2000). Studies of AAI and individuals with ASD demonstrate that AAI may reduce overall symptom severity, stress, problem behaviors, repetitive behaviors, and can improve motor skills and communication (O’Haire, 2017). AAI for individuals with ASD are diverse and can include: working with a service dog, therapeutic horse-back riding, interacting with guinea pigs, and others (Burrows et al., 2016; Gabriels et al., 2015; Grandgeorge et al., 2019). The most researched outcome (or efficacy metric) across AAI is social/communication improvements (O’Haire, 2017). Some researchers believe the human-to-human social engagement difficulties of some individuals with ASD may not carry over to the human-to-animal context (Johnson, 2003).

Historically, the efficacy of AAI has included anecdotal testimonies; however, over the last 15 years the larger field has shifted to more empirical evaluations. For example, research recommendations have shifted to include more stringent research designs, established/validated metrics of social/communication improvement, and best practice guidelines for intervention improvements (Grandin et al., 2015; Palley et al., 2010). The number of empirical AAI and ASD studies published each year has continued to increase from 14 studies total between 1989–2012 (O’Haire, 2013) to 28 more studies published between 2012 and 2015 (O’Haire, 2017). This rapid increase of studies and publications has the potential to set the foundation of a significant evidence base for AAI for individuals with ASD. Though the number of studies is rapidly increasing, their interdisciplinary nature has led to publications in several fields without a synthesis of current findings. Additionally, the quality of the evidence remains unclear. A systematic literature review focused on the state of the evidence is critical for the development of the larger AAI field.

This systematic literature review will synthesize the literature and address the current state of the field of AAI for ASD. The review will replicate previous protocols to seamlessly integrate new findings to our current understanding of the evidence and to shed light on new patterns of rigor and efficacy within this growing body of evidence. The goal of the review is to identify, summarize and evaluate the studies examining AAI for individuals with ASD published between January 2016 and October 2020. Within this review we will: 1) Describe characteristics of the assessed AAIs, 2) Evaluate the methodology, rigor, and risk of bias of each study, 3) Summarize the reported outcomes of AAI for ASD and 4) Compare previous review findings to those of the current review.

Methods

Protocol & Eligibility

Methodology for this systematic literature review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Liberati et al., 2009; Moher et al., 2009). Prior to the start of the study, all study procedures were established (i.e., search strategy, inclusion criteria, data extraction specifications). To build upon prior review findings, inclusion criteria replicated O’Haire (2017), which included:1) publication in English in a peer-reviewed journal, 2) collection of empirical data on AAI or companion animal ownership (empirical defined as the systematic collection and reporting of original observational or experimental scientific research), 3) reporting of outcome results for participants with ASD. Studies were excluded if they were not published in English, literature reviews/meta-analyses, dissertations, validation of a measure, book reviews, book chapters, magazine articles, commentaries, editorials without empirical outcomes, or conference presentations, abstracts or posters.

Literature Search Procedure

Studies analyzed in this systematic literature review were obtained from electronic databases including PubMed, ERIC, Web of Science, ProQuest, PsycINFO and Scopus. Previous systematic literature reviews captured until December 2015, so this review captures studies between January 2016 and October 2020. All database searches included 1) an identifier for ASD (“Autism Spectrum Disorder” or “Autistic Disorder” or “Autism” or “Autistic” and 2) an identifier for AAI or Pet Ownership (see Table 1). The search strategy was adapted to each database by mapping terms appropriately to each database’s vocabulary.

Table 1.

Search Terms: Identifiers for Animal-Assisted Interventions or Pet Ownership

Animal Assisted	Canine therapy	Equine facilitated	Service Animal
Animal Facilitated	Companion Animal	Equine Therapy	Service Dog
Animal therapy	Dog Assisted	Hippotherapy	Service Horse
Anthrozoology	Dog facilitated	Horseback riding	Therapeutic Horseback
Assistance Animal	Dog therapy	Human animal	Therapy animal
Assistance Dog	Dolphin assisted	Human animal bond	Therapy Dog
Assistance horse	Dolphin facilitated	Pet Assisted	Therapy Horse
Canine Assisted	Dolphin therapy	Pet facilitated	Therapy pet
Canine Facilitated	Equine Assisted	Pet Therapy	Therapeutic Riding
			Therapy with animal

Data Extraction and Evaluation

Data was collected via Covidence (Covidence - Better Systematic Review Management, 2020), an online platform that allows for title, abstract, and full text screening with multiple reviewers. Data was imported into Covidence directly from the databases. Two members of the research team screened the titles and abstracts for inclusion into the study (Cohen’s kappa=0.78). The same two members of the team screened full text articles for inclusion (Cohen’s kappa =0.95). Data was extracted from the articles to examine the three aims of the review. Data for Aim 1- describe AAI characteristics which included: AAI terminology, animal species, setting, ratio of interventionist to client, handler certifications, dose (program activities duration, frequency). Data for Aim 2- evaluate the methodology and risk of bias, included: sample size, participant demographics (age, gender, race, ethnicity, diagnosis), study design, effect size, comparison condition, assessment measures, ethical approval, and major limitations. Data for Aim 3- summarize the reported outcomes of AAI for individuals with ASD- included the measures and results of each study. Additional data extracted included: first author, publication year, country, and journal name.

To establish inter-rater reliability for data extraction, the first author and a research assistant independently coded 20% of the articles (84% rater agreement). Methodological rigor and bias was evaluated replicating the procedure outlined in Rodriguez et al., 2020 by extracting items from well-known methodological assessment tools including the National Institutes of Health Study Quality Assessment Tools (LaB NIHNH, 2019), the Consolidated Standards of Reporting Trials (Schulz et al., 2010), the Strengthening the Reporting of Observational Studies in Epidemiology checklists (Von Elm et al., 2014), and the Specialist Unit for Review Evidence Checklists (Specialist Unit, 2019). Summary measures were grouped by most commonly reported outcomes. Additionally, findings from this review were compared with previous literature to acknowledge how this review is unique and adds to the literature as well as how interventions have changed over time.

Results

Database searches identified 306 articles published between January 2016 and October 2020. The PRISMA Flow Chart is presented in Figure 1. After removing duplicates, 140 studies remained for title/abstract screening. Title/abstract screening identified 75 articles that were not relevant to the aims of the systematic review. In this and subsequent sections, n refers to the number of articles. Full text screening of 65 articles identified 22 articles for exclusion due to being a dissertation (n=1), non-empirical (n=6), not published in English (n=6), not autism specific (n=5), not AAI specific (n=2), and solely based on animal outcomes (n=2). The final sample included 43 articles (14% of the initial pool) that met the criteria for inclusion in this systematic review.

Figure 1.

PRISMA Systematic Review Flow Diagram

The number of articles published each year more than doubled between 2016 and 2020; 2016 (n=5), 2017 (n=6), 2018 (n=7), 2019 (n=11), 2020 (n=14). Though articles were only included if they were published in the English language, 6 continents were represented: Africa (n=1), Australia (n=6), Asia (n=2), Europe including Turkey (n=15), North America (n=17) and South America (n=2). Articles were published in a range of journals including those with a veterinary/animal focus (n=6), human animal interaction focus (n=2), human health focus (n=19), autism focus (n=11) and psychology focus (n=5).

Characteristics of AAI for Autism

This section focuses solely on the animal-assisted intervention literature (See Table 2, n=36) and will not include the companion animal specific articles (n=6) and service animal article (n=1).

Table 2.

Overview of Animal-Assisted Intervention (AAI) Characteristics

Reference	N	Terminology	Setting	# of clients, interventionists and animals (per session)	Practitioner Certification	Program Duration (weeks)	Session Frequency (per week)	Session Length (minutes)	Total Contact Time (hours)
HORSES
Petty et al 2017	67	Therapeutic horseback riding	Horse farm	3–4 children, 2 side walkers per child, 1 leader per child, 1 instructor/therapist	Professional Association of Therapeutic Horsemanship (PATH) & Master’s Level Therapist	10 week	1x/week	60 minutes	10 hours
Gabriels et al 2018	64	Therapeutic horseback riding	Horse farm	2–4 children, 2 side walkers per child, 1 leader per child, 1 instructor per group, 1 horse per child	Professional Association of Therapeutic Horsemanship (PATH)	10 weeks	1x/week	45 minutes	7.5 hours
Coman et al 2018	50	Equine assisted activities	Horse farm	1 child, 2 side walkers, 1 leader, 1 instructor, 1 horse	Professional Association of Therapeutic Horsemanship (PATH)	12 weeks	1x/week	60 minutes	12 hours
Souza-Santos et al 2018	45	Equine assisted therapy	Horse farm	1 child, 1 instructor/therapist, 1 horse	Certified therapist	12 weeks	2x/week	60 minutes	24 hours
Kwon et al 2019	29	Therapeutic horseback riding	Horse farm	1 child, 2 side walkers, 1 leader, 1 instructor, 1 horse	Rehabilitation horseback riding instructor with national license with more than 5 years of experience	8 weeks	1x/week	30 minutes	4 hours
Harris et al 2017	26	Horse riding intervention	Horse farm	1participant, 2 side walkers, 1 instructor, 1 horse	British Horse Society (BHS)	5–7 weeks	1x/week	45 minutes	3.8–525 hours
Ozyurt et al 2020	24	Equine assisted activities	Horse farm	2 children, 2 side walkers per child, 1 leader per child, 4 practitioners, 1 horse per child	Professional Association of Therapeutic Horsemanship (PATH), Occupational therapist, Physical therapist, Speech and Language therapist, Pediatrician	8 weeks	1x/week	60 minutes	8 hours
Pan et al 2019	16	Therapeutic horseback riding	Horse farm	2–4 children, 1 volunteer per child, 1 therapeutic riding instructor, 1 horse per child	Professional Association of Therapeutic Horsemanship (PATH)	10 weeks	1x/week	45 minutes	7.5 hours
Anderson et al 2016	15	Equine assisted activities	Horse farm	5 children, 1 instructor, volunteers (# not specified), 5 horses	Riding for the Disabled (RDA) & British Horse Society (BHS) trained instructor	6 weeks	1x/week	180 minutes	18 hours
Bystrom et al 2019	9	Animal-assisted therapy	Horse farm	4–5 children, 6–7 therapists, 1 person responsible for animals, horses (# number not specified)	Psychologist, Physiotherapist, Occupational therapist, Special pedagogue, Teacher, and Teacher’s aide	40 weeks	1x/week	150 minutes	100 hours
Malcolm et al 2018	9	Equine therapy	Horse farm	1 child, 2 side walkers, 1 leader, 1 teacher, 1 horse	Not provided	Not provided	Not provided	60 minutes	not provided
Llambias et al 2016	7	Equine assisted occupational therapy	Horse farm	1 child, 2 side walkers, 1 leader, 1 therapist, 1 horse	Professional Association of Therapeutic Horsemanship (PATH), Occupational therapist	10 weeks	18–23 sessions	45–60 minutes	13.5–23 hours
Peters et al 2020	6	Occupational therapy in an equine environment	Horse farm	1 child, 2 volunteers, 1 therapist, 1 horse	Professional Association of Therapeutic Horsemanship (PATH), Occupational therapist	10 weeks	1x/week	45–60 minutes	7.5–10 hours
Kalmbach et al 2020	5	Occupational therapy in an equine environment	Horse farm	1 child, 1therapist, 1 horse	Occupational therapists	10 weeks	1x/week	45–60 minutes	7.5–10 hours
Tan et al 2018	5	Equine assisted interventions	Horse farm	1 child, sometimes 2 side walkers, 1 leader, 1 instructor, 1 horse	Trained mental health professional	4 weeks	1x/week	not provided	Not provided
De Milander et al 2016	2	Equine assisted therapy	Horse farm	1 client, 1 practitioner, 1 horse	Kinder Kineticist	10 weeks	1x/week	30 minutes	5 hours
Cerino et al 2016	1	Equine assisted interventions	Horse farm	1 child, 1 psychologist, 1 horse	Italian Federation of Equestrian Sports (FISE)	80 weeks	1x/week	60 minutes	80 hours
van der Steen et al 2019	1	Equine assisted interventions	Horse farm	1child, 2 volunteers, 1 educator, 1 horse	Educator & New Trails Learning System equine specialist	5 weeks	1x/week	90–120 minutes	7.5–10 hours
DOGS
Wijker et al 2020 (B)	52	Animal-assisted therapy	Psychiatric outpatient Center	1 client, 1 therapist, 1 dog	Therapist	10 weeks	1x/week	60 minutes	10 hours
Germone et al 2019	47	Animal-assisted activity	Hospital	2–4 children, 1 experimenter, 1 counselor, 1 volunteer handler, 1 dog	Mental health counselor, Volunteer dog handler - registered as a Therapy Team through Pet Partners and evaluated by a hospital volunteer veterinarian	1 session	2x/week	32 minutes	Not Provided
Becker et al 2017	31	Animal-assisted social skills training	School	7–8 children, 2 therapists, 2 dog handlers, 2 dogs	Graduate degree in psychology or social work	12 weeks	1x/week	60 minutes	12 hours
Grandgeorge et al 2017	29	Animal-assisted intervention	School	1 child, 1 trainer, 1 dog, 1 video recorder	Animal trainer (training not specified)	Not provided	1x/week	20–30 minutes	Not provided
Silva et al 2019	23	Animal-assisted intervention	Home or therapy center	1 participant, 2 experimenters (1 for videoing only), 1 dog or robot dog	Certified therapy dogs (ANIMAS)	6 weeks	1x/every other week	Not Provided	Not provided
Hill et al 2020 (B)	22	Animal-assisted occupational therapy	Therapy office	1 child, 1 therapist, 1 therapy dog, 1 parent	Occupational therapist, Trained therapy dog (training not specified)	9 weeks	1x/week	60 minutes	9 hours
Avila-Alvarez et al 2020	19	Animal-assisted intervention	Hospital	1 child, 1 therapist, 1 dog (plus parent or guardian)	Occupational therapist with specialized training in AAI	9 weeks	1x/week	20 minutes	3 hours
London et al 2020	17	Animal-assisted therapy	Not specified	1 child, 1 therapist, 1 dog trainer, 2–3 dogs	Occupational therapist, assistance dog trainer (Assistance Dogs Australia)	5 weeks	1x/week	60 minutes	5 hours
Michelotto et al 2019	15	Animal-assisted activity	Therapy office	11 children, therapists (# not reported),1 researcher, 1 veterinarian, 2 dogs	Trained therapist, Handler certification not specified, dogs were considered suitable for the intervention (calm, confidence, liked play, like touch, emotionally stable, did not reactive to sudden movement)	10 weeks	1x/week	30 minutes	5 hours
Hill et al 2020 (A)	10	Animal-assisted occupational therapy	Therapy office	1 child, 1 therapist, 1 therapy dog, 1 parent	Occupational therapist, Trained therapy dog (training not specified)	9 weeks	1x/week	60 minutes	9 hours
Silva et al 2018	10	Animal-assisted intervention	Home	1 participant, 2 experimenters (1 for videoing only), 1 dog or robot dog	Certified therapy dogs (ANIMAS)	6 weeks	1x/every other week	Not Provided	Not provided
Hill et al 2020 (C)	6	Animal-assisted occupational therapy	Therapy office	Not applicable	Occupational therapist	9 weeks	1x/week	60 minutes	9 hours
Wijker et al 2020 (A)	6	Animal-assisted therapy	Psychiatric outpatient Center	1 client, 1 therapist, 1 dog	Therapist	10 weeks	1x/week	60 minutes	10 hours
Jorgenson et al 2020	5	Therapy animal	Psychiatric outpatient center	1 child, 1 therapist, 1 handler, 1 dog	Therapist, Assistance Dogs International trained dog/handler team	Not provided	Not provided	5 minutes	Not provided
Protopopova et al 2020	5	Therapy animal	Research facility	1 child, 1 researcher, 1 therapy dog handler, 1 dog	Trained handler and dogs team (advanced obedience skills, registered with Pet Partners or Alliance of Therapy Dogs)	19–25 sessions (weeks not specified)	2–3x/week	30 minutes	9.5–12.5 hours
DOLPHINS
Ashtari et al 2018	10	Swimming with dolphins	Dolphin Park	Not reported	Not provided	16 sessions	Not provided	Not provided	Not provided
Griffioen et al 2019 (A)	5	Animal-assisted therapy	Dolphin Park	1 child, 1 therapist, 1 dolphin trainer, 1 dolphin	Therapist, Trainer	6 weeks	1x/week	Not provided	Not provided
MUTIPLE SPECIES
Kregiel et al 2019	50	Animal-assisted therapy	Varied among participants	Varied among participants	Varied among participants	Varied	Varied	Varied	Varied

Note: Table is sorted by descending sample size within each species category.

Terminology

The majority of the articles used “assisted” to describe the intervention (72%, n=26). The next most common term used was “therapeutic horseback riding” (12%, n=4). Other terms included “horse-riding intervention” (2%, n=1), “occupational therapy in an equine environment” (6%, n=2), “swimming with dolphins” (2%, n=1) and “therapy dog” (6%, n=2).

Species

The most common animals in AAI for ASD were horses (50%, n=18), followed by dogs (42%, n=15), dolphins (6%, n=2) and multiple species together (2%, n=1). One article focused on animal-assisted interventions that mentioned multiple species (Kregiel et al., 2019) and examined 50 parent reports of their child’s behavior and motor activity comparing results across multiple forms of AAIs including canine-assisted therapy (40% of sample), equine-assisted therapy (38% of sample), feline-assisted therapy (12% of sample), and dolphin-assisted therapy (10% of sample). All but two articles mentioned credentials of the interventionists (94%, n=34) and multiple articles included multiple credentials. The most common certification was Professional Association of Therapeutic Horsemanship certification (19%, n=7). Other certifications included Riding for the Disabled (2%, n=1), certifications from the British Horse Society (6%, n=2), certifications from Italian Federation of Equestrian Sports (FISE) (2%, n=1), and national (Korea) rehabilitation horseback riding licensure (2%, n=1). Aside from certifications, credentials included post-secondary schooling (47%, n=17) and training specific for the target AAI (42%, n=15). None of the equine-assisted intervention studies mentioned training/certification specific to the equine. Five canine-assisted intervention studies mentioned that the dogs were trained but did not specify the organization. Ten canine-assisted intervention studies specified an organization that had trained and/or certified the canine used in the intervention (e.g. Pet Partners, Assistance Dogs International, Alliance of Therapy Dogs).

Ratio of Interventionist to Client to Animal

The majority of studies shared the ratios of clients/interventionists and/or clients/animals (86%, n=31). The most common ratio for equine-assisted interventions was 1:1:2:1 (25%, n=9, one child riding one horse with two side walkers next to the horse and one leader leading the horse). This structure often was replicated with multiple children in the group with one therapeutic riding instructor and follows the Professional Association of Therapeutic Horsemanship guidelines (one child, one horse, two side walkers, one horse leader, and one therapeutic riding instructor). The most common ratio for canine-assisted interventions was one client to one therapist working with their own dog (14%, n=5). Only one of two dolphin-assisted studies clearly stated the number of interventionists to clients.

Dose

The program duration, frequency and session length were collected to obtain total contact time. Total program duration ranged from one session to 80 sessions. The average duration (number of sessions) was 12.36 sessions. Most studies reported that interventions took place weekly (72%, n=26). Other studies reported multiple times per week (6%, n=2), one session every other week (6%, n=2) and others did not specify frequency (17%, n=6). Length of session ranged from 20 minutes to three hours with an average of about one hour (83%, n=30). The total contact time of the intervention ranged from 3–100 hours with 15.7 hours on average (72%, n=26).

Methodology

All studies (service animal, companion animal and AAI programs) included in the review were analyzed for methodological rigor and risk of bias (Appendix 1, Figure 2, n=43). All 43 manuscripts clearly stated the objective of the study, but only 28% (12/43) stated a directional hypothesis. Institutional review board approval was mentioned in 91% (39/43) of studies. Most of the studies shared participant demographic information (91%, 39/43), diagnosis of participants (100%, 43/43) and inclusion/exclusion criteria (79%, 34/43). Only 21% (9/43) of studies demonstrated that the groups were equal in terms of demographic variables. Variability for outcomes and specific statistical values (t, F or B) were shared in 60% (26/43) of studies. Precise p-values were shared in 58% (25/43) of studies. Effect sizes were shared in 39% (17/43) studies and 81% (35/43) of studies mentioned at least two limitations of the study design. The methodological rigor of individual studies ranged from meeting 33% to 100% of the methodological considerations assessed (average = 76%, Appendix 1).

Figure 2.

Methodological Rigor

Sample

Participants were both individuals with ASD (n=30) as well as parents/caregivers (n=12) and therapists (n=1). Sample size ranged from one participant (case study) to 764 participants (online survey).

Participants - Individuals with ASD

The average sample size of participants with ASD was 23.22 with a range of 1–67 individuals. Roughly half (52%) of studies assessing individuals with ASD had sample sizes less than 20. The ages of individuals with ASD ranged from two years to 60 years with an average age of 9.57 years and 80% of study participants were male. Six studies (14%) reported participant race/ethnicity and 72% of participants in those studies were white.

Participants - Parents/Caregivers

The average sample size of studies of parents and caregivers was 121.2 with a range of 5–764 individuals. Half of the parent/caregiver focused studies had sample sizes less than 20 individuals. Most (89%) of the reporting parents/caregivers were female. Only 25% (n=3) of studies of parents/caregivers reported race/ethnicity of participants with 65% of participants endorsing White as their racial classification. One study interviewed six female occupational therapists as the study participants to capture their perspective on the intervention (Hill et al., 2020c).

Design and Effect Size

Both qualitative and quantitative methodologies were present in the review (Table 3). Most of the articles were quantitative (77%, n=33) with about 23% being qualitative papers (n=10). Within the quantitative papers, 12 were non-randomized control, seven used pre/post, eight used randomized control, four used single subject approaches, and two used cross-sectional surveys. A little more than one-third of the studies (n=17, 39%) reported effect size.

Table 3.

Research Study Design and Assessments

Reference	N	Study Design	Effect size	Survey	Interview	Observation	Physiology	Independent Researcher Rater	Therapist Rater	Parent/Caregiver Rater	Teacher Rater	Self Rater	Blinded	IRB	IACUC	Total Bias (%)
HORSES
Petty et al 2017	67	Randomized Control	X	X						X				X		100%
Gabriels et al 2018	64	Randomized Control	X	X					X	X				X		92%
Coman et al 2018	50	Randomized Control	X	X						X	X1					85%
Souza-Santos et al 2018	45	Randomized Control		X				X						X		92%
Kwon et al 2019	29	Non-Randomized Control		X					X					X		92%
Harris et al 2017	26	Non-Randomized Control	X	X		X					X1			X		92%
Ozyurt et al 2020	24	Randomized Control	X	X				X		X				X		92%
Pan et al 2019	16	Randomized Control	X	X		X	X	X	X	X				X		92%
Anderson et al 2016	15	Pre/Post	X	X	X					X				X		83%
Bystrom et al 2019	9	Qualitative				X			X					X		57%
Malcolm et al 2018	9	Qualitative			X					X	X2			X		43%
Llambias et al 2016	7	Single Subject	X			X			X				X	X		73%
Tan et al 2018	7	Qualitative			X					X				X		86%
Peters et al 2020	6	Single subject	X	X		X				X				X		91%
Kalmbach et al 2020	5	Qualitative			X					X				X		86%
De Milander et al 2016	2	Pre/Post		X		X			X					X		55%
Cerino et al 2016	1	Qualitative				X			X							43%
van der Steen et al 2019	1	Pre/Post		X	X	X		X		X				X		33%
DOGS
Carlisle et al 2018	338	Qualitative		X						X				X		71%
Fecteau et al 2017	98	Non-Randomized Control	X	X			X			X						92%
Wijker et al 2020 (B)	52	Randomized Control	X	X						X3		X		X		77%
Germone et al 2019	47	Non-Randomized Control				X		X						X		92%
Hall et al 2016	37	Non-Randomized Control	X	X						X				X		85%
Becker et al 2017	31	Non-Randomized Control	X	X					X		X1	X	X	X	X	100%
Grandgeorge et al 2017	29	Non-Randomized Control				X		X					X	X	X	69%
Silva et al 2019	23	Non-Randomized Control	X			X	X	X						X	X	100%
Hill et al 2020 (B)	22	Randomized Control		X		X		X		X				X	X	77%
Avila-Alvarez et al 2020	19	Pre/Post	X	X					X					X		83%
London et al 2020	17	Qualitative			X					X				X		86%
Michelotto et al 2019	15	Pre/Post		X					X	X				X		58%
Harwood et al 2019	11	Qualitative			X						X1			X		71%
Hill et al 2020 (A)	10	Qualitative			X					X				X		86%
Silva et al 2018	10	Non-Randomized Control	X			X	X	X						X		100%
Hill et al 2020 (C)	6	Qualitative			X				X					X		86%
Wijker et al 2020 (A)	6	Pre/Post		X		X		X				X		X		67%
Jorgenson et al 2020	5	Single Subject		X		X		X								45%
Protopopova et al 2020	5	Single Subject				X	X	X						X	X	64%
DOLPHINS
Ashtari et al 2018	10	Non-Randomized Control		X				X						X		62%
Griffioen et al 2019	5	Pre/Post				X		X						X		50%
CATS
Hart et al 2018	108	Non-Randomized Control		X	X					X			X	X		50%
MUTIPLE SPECIES
Carlisle et al 2020	764	Cross-Sectional	X	X						X				X		77%
Kregiel et al 2019	50	Cross-Sectional		X						X				X		58%
Grandgeorge et al 2020	42	Non-Randomized Control		X		X		X						X		85%

Note: Table is sorted by descending sample size within each species category. IRB: Institutional Review Board, IACUC: Institutional Animal Care and Use Committee

¹School Teacher,

²Teachers/Staff at Equine Center,

³Informant

Control/Comparison

Only 46% (20/43) studies included a comparison/control group, while nine studies did not have a comparison/control. In 14 studies a comparison/control was not applicable. The most common control was the intervention without an animal present (16%, n=7) such as a barn activity without a live horse (7%, n=3), a no dog present therapy group (5%, n=2), a no companion animal in the home group (n=1) or a water activity without a dolphin (2%, n=1). Multiple studies used an alternative item (12%, n=5) in place of the live animal. These items included a robotic dog (5%, n=2), a toy (2%, n=1), or a preferred leisure item identified by the parent or child (2%, n=1). Other studies used a waitlist control (9%, n=4) or usual care (5%, n=2). Two studies compared children with ASD to typically developing children (5%, n=2).

Assessment

The most common type of assessment used in the studies were surveys (63%, n=27) including both standardized and non-standardized survey measures. The second most common type of assessment was observation (42%, n=18) followed by interviews (23%, n=10) and physiology (salivary cortisol 7%, n=3, heart rate 5%, n=2). Most studies (65%, n=28) used one form of assessment (e.g. survey, interview, observation, physiology) followed by 35% using multimodal assessment (n=15). The majority of studies (51%) used parents/caregivers as raters (n=22). The second most common rater was independent research raters (35%, n=15) followed by therapists as raters (26%, n=11). Five studies (12%) included teachers (four studies used school teachers, one study used the teachers/staff at the equine center) as raters and three studies (7%) used self-rating. Only four studies (9%) incorporated blind-to-condition raters to reduce the risk of bias.

Ethical Review

Thirty-nine studies (91%) reported an ethical approval process (e.g. Institutional Review Board) for the human subjects. In comparison, only five studies (11%) reported an ethical approval process for the animals within the intervention (e.g., Institutional Animal Care and Use Committee). All the studies that reported an ethical approval process for animals were canine-assisted interventions (11%, n=5).

Outcomes

All studies in this review (n=43) were evaluated for outcomes of AAI on ASD (see Table 4).

Table 4.

Comparison Conditions, Measures, and Outcomes

Reference	N	Control or Comparison Condition	Standardized Outcome Measures or Qualitative	Animal-Assisted Intervention (AAI) Outcomes
HORSES
Petty et al 2017	67	Barn activity without horse involved	Child’s Attitude and Behavior toward Animals (CABTA)	+/ CABTA: + Animal Attachment Scale, / Animal Abuse Score
Gabriels et al 2018	64	No Horse Barn Activity	Aberrant Behavior Checklist-Community (ABC-C), Social Responsiveness Scale (SRS), Systematic Analysis of Language Transcripts (SALT)	+/ ABC-C: + Irritability, / Hyperactivity + SRS: + Social Communication, + Social Cognition + SALT
Coman et al 2018	50	Waitlist Control	Social Responsiveness Scale (SRS), Sensory Profile (SP), Sensory Profile School Companion (SPSC)	+/ SRS Parent: + Total Score, / Social Awareness, + Social Cognition, + Social Communication, + Social Motivation, + Autistic Mannerisms + SRS Teacher: + Total Score, + Social Awareness, + Social Cognition, + Social Communication, + Social Motivation, + Autistic Mannerisms +/ SP Parent: + Total Score, + Emotionally Reactive, + Low Endurance Tone, + Inattention/Distractibility, + Sedentary, / Sensory Sensitivity, / Sensation Seeking, / Oral Sensory Seeking, / Poor Registration, / Fine Motor Perception + SPSC Teacher: + Registration, + Seeking, + Sensitivity, + Avoiding
Souza-Santos et al 2018	45	Dance Intervention	Childhood Autism Rate Scale (CARS), Functional Independence Measure (FIM), WHO Disability Assessment Scale (WHODAS 2.0, social participation scale)	+ CARS / FIM / WHODAS 2.0 (EAT intervention only)
Kwon et al 2019	29	Conventional therapy	Receptive and Expressive Vocabulary Test (REVT) and Preschool Receptive-Expressive Language Scale (PRES) and Kaufman Assessment Battery for Children II (K-ABC-II) or cognitive domain of Bayley Scales of Infant Development II (BSID-II)	/ REVT: / Reception, / Expression / PRES: / Reception, / Expression / Cognitive domain BSID-II / K-ABC
Harris et al 2017	26	Education as usual	Childhood Autism Rating Scale (CARS-2), Aberrant Behavior Checklist- Community (ABC-C), Measurement of Pet Intervention Checklist (MOPI)	/ CARS-2 / ABC-C: / Hyperactivity, / Irritability / MOPI
Ozyurt et al 2020	24	Normal Care	Children’s global assessment scale (CGAS), McMaster Family Assessment Device (FAD), social communication questionnaire (SCQ), Beck Depression Inventory (BDI)	+ CGAS +/ FAD: / Problem solving, + Communication, + Roles, / Affective responsiveness, + Affective involvement, + Behavioral control, / General functions + SCQ + BDI
Pan et al 2019	16	Barn activity without horse involved	Systematic Analysis of Language Transcripts (SALT), Social Responsiveness Scale (SRS), Aberrant Behavior Checklist-Community (ABC-C), Salivary Cortisol	+ ABC-C: +Hyperactivity, + Irritability, / Lethargy/Social Withdrawal, / Stereotypy, / Inappropriate Speech Behaviors +/ SRS: + Social Awareness, / Social Cognition, / Social Motivation, / Social Communication, / Autistic Mannerisms / SALT / Cortisol
Anderson et al 2016	15	None (Pre/post)	Autism Spectrum Quotient (ASQ), Vineland Adaptive Behavior Scale (VABS), Empathizing Quotient (EQ), Systemizing Quotient (SQ)	+ ASQ +/ VABS: + Maladaptive Behavior, / Adaptive Behavior, / Communication, / Socialization +/ EQ/SQ: + Empathizing Quotient, / Systemizing Quotient
Bystrom et al 2019	9	N/A	Qualitative	+ Reduction of stress, Increased calmness and Peaceful state of mind, Increased curiosity and interest, Increased likelihood of spontaneous attention
Malcolm et al 2018	9	N/A	Qualitative	+ Improved social interaction and communication + Embodied multisensory experience
Llambias et al 2016	7	N/A	Engagement (video coding)	+ Engagement
Peters et al 2020	6	N/A	Visual analog scale, Aberrant Behavior Checklist- Community (ABC-C) Hyperactivity and Irritability Subscales, Social Responsiveness Scale (SRS-2)	+ ABC-C: + Hyperactivity (4/5 participants), +Irritability (4/5 participants) +/ SRS-2: + Social motivation , + Social communication, / Social awareness, / Social cognition , / Restricted interests and Repetitive behaviors
Kalmbach et al 2020	5	N/A	Qualitative	+ Parental perspectives on child’s experiences of the intervention (qualities of the horse and occupational therapist that favorably contributed to child’s experiences & children’s positive emotional experiences regarding the intervention) + Parental perspectives on the intervention’s influences on everyday life (favorable influences on child’s everyday life, favorable influences on family’s everyday life) (−) Parent concerns and dissatisfaction (e.g. too rigid, too limited, travel times)
Tan et al 2018	5	N/A	Qualitative	+ Improved self-concept and enhanced emotional well-being + Child’s improved self-regulatory ability + Social benefits for the child + Benefits to Parents
De Milander et al 2016	2	None (Pre/post)	Bruininks-Oseretsky Test of Motor Proficiency (BOT-2)	+ BOT-2: + Balance, + Coordination, + Strength
Cerino et al 2016	1	N/A	Qualitative	+ Relationship with animal was used to encourage narrative from child to improve cognition and communication and relationship to the therapist. + Increased presence due to the safety created in the physical environment.
van der Steen et al 2019	1	None (Pre/post)	Scale for Emotional Development-Revised (SED-R), Strengths and Difficulties Questionnaire (SDQ), Observed Social and Communication Skills	Parents: +/ SED-R: / Self-image in interacting with environment, / Anxieties, / Interacting with Peer, / Handling Materials, / Play Development, + Interacting with emotionally important others, + dealing with a changing environment, + Communication, + Aggression Regulation, + Moral Development, + Deal with her own body, + Emotion differentiation, + Emotion regulation + SDQ Total Difficulties +/ Social and Communication Skills: + Peer Relationship Problems (both mother & father), / Pro Social Behavior (mother not father), + Hyperactivity/inattention (both mother & father), / Conduct problems (both mother & father), / Emotional symptoms (mother not father) Researcher: + Verbal Communication, / Tension, + Turn-taking, + Separation Anxiety, + Positive emotion, / Negative emotions
DOGS
Carlisle et al 2018	338	N/A	Qualitative	+ Theme 1: Facilitator and Benefit , + Theme 2: Learning Opportunity (−) Theme 3: Barrier, (−) Theme 6: Safety / Theme 4: Grief, / Theme 5: Fit, / Theme 7: Alternative Animal Options
Fecteau et al 2017	98	Waitlist Control group	The Parenting Stress Index - short form (PSI-SF), Salivary Cortisol	+/ PSI-SF: + Total Score, / Parental Distress Scale, + Parent-Child dysfunctional interaction scale, + Difficult Child Scale + Wakening and Morning Cortisol
Wijker et al 2020 (B)	52	Waitlist Control	Percieved stress scale (PSS), symptom checklist-90-revised (SCL-90-R), social responsiveness scale of adults(SRS-A), Rosenberg Self-Esteem Scale (RSES)	/ PSS / SRS (rated by informant), / SRS (rated by self) / SCL-90- R / RSES
Germone et al 2019	47	Toy	Observation of Human-Animal Interaction for Research tool (OHAIRE)	+/ OHAIRE: + Communication, + Interaction, + Expression, / Interference behavior
Hall et al 2016	37	No Dog	Parenting Stress Index-Short Form (PSI-SF), Family Functioning (Brief FAM-III-GS)	+ FAM-III-GS / PSI-SF: / Total Score, / Parental Distress Scale, / Parent-Child dysfunctional interaction scale, / Difficult Child Scale
Becker et al 2017	31	Social Skills Group without the Dog	Social Responsiveness Scale (SRS-2), Children’s Depression Inventory (CDI-2), Social Language Development Test (SLDT), Reading the Mind in the Eyes Test (RMET)	+ SRS-2: + Social communication & interaction subscale, + Repetitive behavior subscale, + Total score +/ CDI-2: + Total score, / ineffectiveness scale, + functional problems subscale, / Negative Mood/Physical Symptoms, / Negative Self Esteem, / Emotional Problems, + Interpersonal Problems / SLDT: / Making Inferences, / Supporting Peers / RMET
Grandgeorge et al 2017	29	No Dog	Video Recording	+ Visual Attention, + Orientation, + Joint Attention
Silva et al 2019	23	Robotic Dog, No Stimulus	Heart Rate Variability (LnRR, LnSDNN), Social Communication Behavior Coding	+ Social Communication + Heart Rate Variability: + LnRR, + LnSDNN
Hill et al 2020 (B)	22	Usual Care Occupational Therapy	Video Coding, Canadian occupational performance measure (COPM)	/ On task behavior / COPM performance and satisfaction
Avila-Alvarez et al 2020	19	None (Pre/post)	Assessment of Communication and Interaction Skills (ACIS), Animal-assisted Therapy Flow Sheet (Richeson & McCullough, 2002)	+ ACIS Total Score (12/20 subscales significant) + Animal-assisted therapy flow sheet (6/9 items significant)
London et al 2020	17	N/A	Qualitative	+ Overarching Theme: Engagement as a precursor to progress + Characteristics of the dog + Therapeutic context + Goal directed achievements
Michelotto et al 2019	15	None (Pre/post)	Autistic Behavior Assessing Questionnaire (CACS-27)	Therapist Perception: −/ Environmental Factors −+/ Intrinsic Factors Parents Perception: −+/ Environmental Factors −+/ Intrinsic Factors
Harwood et al 2019	11	N/A	Qualitative	+ Theme 1: Love and Companionship + Theme 2: Perception of Ownership + Theme 3: Comfort and Calming Influence + Theme 4: Canine’s Ability to Assist Child in Understanding Their World (−) Theme 5: Challenging Experiences
Hill et al 2020 (A)	10	N/A	Qualitative	+ Dog provides emotional safety and helped to build rapport between therapist and child / The dog is not sufficient for therapy engagement, the qualities of the therapist and goal directed nature of the intervention matter significantly.
Silva et al 2018	10	Robotic Dog, Toy	Heart Rate and Heart Rate Variability	+ Heart rate / Heart Rate Variability Parameters
Hill et al 2020 (C)	6	N/A	Qualitative	+ Incorporation of therapy dog in the session accelerates motivation in the therapeutic process (−) identified challenges
Wijker et al 2020 (A)	6	None (Pre/post)	Social Responsiveness Scale for Adults (SRS-A) subscales social awareness, social communication and social motivation, Rosenberg self-esteem scale (RSES)	/ SRS- A: / Social Awareness, / Social Communication, / Social Motivation / RSES
Jorgenson et al 2020	5	N/A	Multiple Stimulus Without Replacement Preference Assessment (MSWO)	/ Noncontingent access to therapy dog slightly increased verbal statements for 1/5 participants / Contingent access to therapy dog increased social interactions for 2/5 participants
Protopopova et al 2020	5	N/A	Salivary Cortisol and Preference Assessment	+ Decrease in cortisol (4/5 participants in dog conditions) + Preference Assessment ( For 4/5 children the dog functioned as a reinforcer)
DOLPHINS
Ashtari et al 2018	10	No Dolphin Water Activity	2 selected subscales of the Bruininks-Oseretsky test for motor proficiency	+ Bruininks-Oseretsky test: + Balance + Strength
Griffioen et al 2019 (A)	5	None (Pre/post)	Video Recording	+/ Turn taking: + (2/5 participants), / (3/5 participants) +/ Spoken language: + (3/5 participants), / (2/5 participants)
CATS
Hart et al 2018	108	Typically Developing Children	Non-Standardized Survey	/ Aggression in cats not heightened with children with ASD. / Cats less affectionate with children with ASD but still moderately affectionate. + Valuable bonding, attention and calming affection to the child
MUTIPLE SPECIES
Carlisle et al 2020	764	None	Lexington Attachment to Pets Scale (LAPS), Companion Animal Bonding Scale (CABS); Animal Ownership for Families of Autistic Children Scale (AOFACS); Parental Stress Scale (PSS)	+ 90% of parents and 71% of children said they were attached or very attached to companion animals −/ 4% of parents said child was “always or “very often” and 22% “sometimes” bothered by the pet −/ 1% of parents said child “always” or “very often” and 9% “sometimes” harmed their animal −/ 13% of parents said child was afraid of companion animal + Moderators of human-animal bond: + Income +/ Moderators of burden: + Stress, / Education, / Income
Kregiel et al 2019	50	None	Non-Standardized Survey	+ 88% of parents say children had more frequent expression of feelings and emotions + 98% says increase in lively gestures + 86% more frequent vocal and verbal reactions + 52% increased responsibility and independence (−) limited access to services
Grandgeorge et al 2020	42	Typically developing children	Observation	/ Increased visual attention in children with ASD with pet cats then pet dogs

Note: Total score is indicated for scales unless subscales are listed. Table is sorted by descending sample size within each species category.

ASD Symptom Severity

A total of six studies evaluated autism severity. Qualitatively, one study explained overall improvement in ASD symptom severity (Malcolm et al., 2018). A total of five different standardized measures were used to evaluate ASD symptom severity as an outcome measure. A significant reduction of ASD symptoms was demonstrated with the Autism Spectrum Quotient (n=1), the Childhood Autism Rating Scale (n=2), the Aberrant Behavior Checklist (n=1), and the Children’s Global Assessment Scale (n=1) (Anderson & Meints, 2016; Harris & Williams, 2017; Ozyurt et al., 2020; Souza-Santos et al., 2018). However, Wijker et al., 2020b used the Symptom Checklist-90-Revised and reported no significant change in symptoms.

Only two studies documented outcomes for restricted and repetitive behaviors (Michelotto et al., 2019; Peters et al., 2020). One study used the Aberrant Behavior Checklist-Community and reported no significant changes in repetitive behaviors (Peters et al., 2020). Another study used the Autistic Behavior Assessing Questionnaire and reported the opposite in that therapist and parent raters perceived reduction in repetitive stereotyped movements and rituals after AAI (Michelotto et al., 2019).

Social Interaction

Social interaction as the most common outcome evaluated within the studies of this review. Qualitative outcomes suggested that AAI can improve social interaction for some individuals with ASD (n=7). AAI increased joint attention and overall social skills (Grandgeorge et al., 2017; Kalmbach et al., 2020), increased the number and quality of social interactions (Jorgenson et al., 2020; Tan & Simmonds, 2018), and the number of appropriate social responses (Llambias et al., 2016; Silva et al., 2019). Additionally, the embodied interactions inherent in equine-assisted interventions built skills of empathy, love and positive affection (Carlisle et al., 2018; Malcolm et al., 2018).

A total of nine different standardized measures were used to evaluate social interaction within the studies of this review. The most commonly used standardized measure was the Social Responsiveness Scale (SRS; n=7). Five studies reported SRS score reductions, indicating less severity of social impairment (Becker et al., 2017; Coman et al., 2018; Gabriels et al., 2018; Pan et al., 2019; Peters et al., 2020). One study reported no significant changes (Wijker et al., 2020a) and another reported significant change when rated by an informant but not with the self-rated SRS (Wijker et al., 2020b). The other measures used to document social improvements included the Scale for Emotional Development-Revised (n=1, van der Steen et al., 2019), the empathizing quotient (n=1, Anderson & Meints, 2016), the social functioning subscales of Childhood Autism Rating Scale (n=1, Harris et al., 2017) the Aberrant Behavior Checklist- Community Edition (n=1, Harris et al., 2017), the Reading the Mind in the Eyes Test (n=1, Becker et al., 2017), and the peer interactions subscale of the Autistic Behavior Assessing Questionnaire (n=1, Michelotto et al., 2019). Two studies reported no social improvements using two separate measures the Vineland Adaptive Behavior Scale and the WHO Disability Assessment Scale (Anderson & Meints, 2016; Souza-Santos et al., 2018).

Language & Communication

A total of 14 studies reported outcomes related to language and communication. Qualitatively three studies reported improvements in language and communication (Griffioen et al., 2019; London et al., 2020; Malcolm et al., 2018). Observational coding with the OHAIRE coding tool (O’Haire et al., 2015) was used in one study which reported significant between group differences for communication, interaction, and expression (Germone et al., 2019). A total of 12 standardized measures were used to evaluate language and communication. Significant improvements were reported in studies that used the Receptive and Expressive Vocabulary Test (n=1, Kwon et al., 2019), the Preschool Receptive-Expressive Language Scale (n=1, Kwon et al., 2019), the Bayley Scale of Infant Development (n=1, Kwon et al., 2019), the Kaufman Assessment Battery for Children (n=1, Kwon et al., 2019), the Social Communication Questionnaire (n=1, Ozyurt et al., 2020), the SRS (n=2, Pan et al., 2019; Peters et al., 2020), the Scale for Emotional Development-Revised (n=1, van der Steen et al., 2019), the Assessment of Communication and Interaction Skills (n=1, Avila-Alvarez et al., 2020), and the Autistic Behavior Assessing Questionnaire (n=1, Michelotto et al., 2019). No significant changes were reported using the Social Language Development Test (n=1, Becker et al., 2017) or in the Vineland Adaptive Behavior Scale- Communication Subscale (n=1, Anderson & Meints, 2016). For studies that utilized a systematic analysis of language transcripts, mixed results were reported. One study reported significant improvement six months post AAI and another study reported no significant changes directly after AAI (Gabriels et al., 2018; Pan et al., 2019). One study used a non standardized survey measure that indicated improvement to language and communication after AAI (Kregiel et al., 2019).

Problem Behaviors

A total of 11 studies reported findings regarding problem behaviors defined as maladaptive or challenging behaviors. Qualitative reports suggest problem behaviors improved with AAI (n=2, Protopopova et al., 2020; Tan & Simmonds, 2018). A total of eight standardized measures were used to assess problem behaviors. Only two measures were used to assess problem behavior in more than one study, the Aberrant Behavior Checklist-Community Hyperactivity and Irritability subscales (n=2, Pan et al., 2019; Gabriels et al., 2018). The Irritability subscale had significant improvement in both studies. The Hyperactivity subscale only demonstrated significant improvement in one of the two studies. The other six measures included the Vineland Adaptive Behavior Scale- Maladaptive Behavior Subscale (n=1, Anderson & Meints, 2016), the Sensory Profile (n=1, Coman et al., 2018), the Child’s Attitude and Behavior toward Animals (n=1, Petty et al., 2017), the Scale for Emotional Development-Revised (n=1,van der Steen et al., 2019), the Children’s Depression Inventory (n=1,Becker et al., 2017), and the Autistic Behavior Assessing Questionnaire (n=1, Michelotto et al., 2019) all of which documented significant reductions in parent-reported problem behaviors.

Positive Emotion

A total of six studies in this review focused on positive emotion as an outcome of AAI for individuals with ASD. Qualitatively, AAI enhanced emotional well-being, emotional regulation, and positive emotional expression (London et al., 2020; Tan & Simmonds, 2018). Standardized measures included the Scale for Emotional Development-Revised (n=1, significant improvement, van der Steen et al., 2019) and the Rosenberg Self Esteem Scale (n=2, no significant change, Wijker et al., 2020b; Wijker et al., 2020a). One study using a cross sectional survey also noted more positive emotion with AAI (Kregiel et al., 2019).

Motor Skills

A total of three studies focused on motor skills as an outcome of AAI for individuals with ASD. Qualitatively, one study reported that a companion dog facilitated physical activity (Carlisle et al., 2018). Two studies used the standardized Bruininks-Oseretsky Test of Motor Proficiency documenting improvement in coordination, balance and strength (De Milander et al., 2016) and significant increase in gross motor skills (Ashtari & Sheikh, 2018).

Stress

A total of nine studies reported stress as an outcome for AAI. Qualitatively, six studies demonstrated that individuals with ASD experience relief from stress, comfort and calmness when interacting with companion animals or animal-assisted interventions (Byström et al., 2019; Carlisle et al., 2018; Hart et al., 2018; Harwood et al., 2019; London et al., 2020; Silva et al., 2018). Additionally, two studies reported decreased heart rate reactivity when individuals with ASD interacted with a live dog in comparison to a robot dog or a toy (Silva et al., 2018, 2019). Another study reported significant improvement on the Perceived Stress Scale (Wijker et al., 2020b).

Negative Outcomes for Individuals with ASD

In addition to the positive outcomes associated with companion animals and AAI for individuals with ASD, two studies documented negative outcomes. These included grief over the loss of a companion animal, safety concerns for children and animals, and lack of connection between individual and animals (Carlisle et al., 2018; Harwood et al., 2019).

Outcomes- For Parents and Family

A total of seven studies focused on outcomes related to parents and families of children with ASD. Qualitative findings suggest both positive and negative outcomes for the parents and family (n=4). Positive outcomes included favorable influences on family’s everyday life (Kalmbach et al., 2020) and benefits to the parents themselves (Tan & Simmonds, 2018). Negative outcomes included additional burden on parents due to animal welfare issues (Harwood et al., 2019). One study measured physiological impacts of the service dog on parents of an individual with ASD through salivary cortisol and reported altered wakening and morning cortisol profiles with the presence of the service dog (Fecteau et al., 2017).

The most commonly reported quantitative measure was parental stress as measured by the Parenting Stress Index Short Form (PSI-SF) (n=3). One study (Fecteau et al., 2017) reported a significant decrease in the parental perception of stress and total stress in the experimental group (families with an ASD service dog). Another study reported that the perceived burden of a companion animal in the home was correlated with higher parental stress while the perceived benefits of a companion animal were correlated with lower parental stress (Carlisle et al., 2020). The third study also used the PSI-SF but reported no significant change (Hall et al., 2016).

Another measure used by multiple studies for psychological outcomes was the Lexington Attachment to Pets Scale (LAPS). One study proposed that as the relationship between parent and child decreases, the relationship between the parent and the companion dog may increase (Hall et al., 2016). A second study reported that parents with high stress levels have higher LAPS scores indicated a stronger bond with the companion animal (Carlisle et al., 2020). Family functioning was also measured using the Brief FAM-III-GS in Hall et al. (2016). They reported significant improvements to family functioning in the intervention group compared to the control group (those without a companion dog in the home). A decrease in maternal Beck Depression Inventory (BDI) scores and improvement on the McMaster Family Assessment Device (FAD) subscales of affective involvement and behavioral control was also reported (Ozyurt et al., 2020). There were no noted changes on the FAD subscales of problem-solving, affective responsiveness and general functions (Ozyurt et al., 2020).

Outcomes for the Intervention Process

A total of nine studies commented on the process of the animal-assisted interventions. Less than half (n=4) of the studies focused on the importance of the animal within the therapeutic process. Studies suggested that the animals helped to motivate clients to participate in the therapeutic process and that the behaviors of the animal were critical to achieve success of the intervention (Hill et al., 2020c; Llambias et al., 2016; London et al., 2020; Malcolm et al., 2018). Additionally, studies (n=2) suggest that the animal is a mediator for the therapeutic relationship (Ávila‐Álvarez et al., 2020; Cerino et al., 2016). One study spoke about the importance of the therapist, suggesting that the animal alone was not sufficient for engagement in the therapeutic process (Hill et al., 2020a). Another study compared visual attention differences between pet cats and pet dogs suggesting that children with ASD demonstrated a stronger connection with their cats (Grandgeorge et al., 2020). Parents also shared concerns about the intervention – child improvements were mostly on or near intervention days, the length and frequency of the intervention was limiting, the delivery of sessions was too rigid, it was expensive and travel was difficult (Kalmbach et al., 2020).

Discussion

In the last five years the literature on ASD and AAI has doubled in size from 42 studies in 2015 to 85 studies (cumulative total assessed in 2020). The goal of this review was to identify and evaluate the studies published from 2016 onwards and to integrate these findings to identify how the evidence base of AAI for ASD has improved or evolved. There were four objectives for the current review 1) Describe characteristics of AAI for ASD, 2) Evaluate the methodology and risk of bias of each study, 3) Summarize the reported outcomes of AAI for ASD and 4) Compare previous review findings to those of the current review within each objective. This review builds on the previous literature by highlighting areas where growth has occurred as well as areas that need critical improvement before a solid foundation of evidence is apparent.

Characteristics of Animal-Assisted Interventions for Autism Spectrum Disorder

Findings suggest that there is great heterogeneity in AAI for ASD. This heterogeneity is not unlike the heterogeneity in the field of psychotherapy (Kazdin, 2017), but it has yet to be appropriately addressed within industry and research protocols. Many researchers highlight this need and recommend strategic plans to improve the evidence base for AAI (e.g., Kazdin, 2017) or to establish and test specific protocols for AAI and ASD (e.g., O’Haire, 2017), but this has yet to occur. However, findings do suggest a significant increase in the use of the recommended terminology (i.e., animal-assisted, equine-assisted) in research published over the last five years which is important to promote clarity across the field and literature (Fine et al., 2015).

In addition to recognizing and addressing the heterogeneity of the intervention itself, future study designs should recognize and incorporate metrics of individual difference to capture variability present in ASD as a “spectrum” disorder. Future studies should also address treatment moderators and predictors in addition to efficacy of the intervention. With a lack of research in this area and the variability in the needs of individuals with ASD (e.g., sensory preferences, communication, symptom severity), it is difficult to personalize a treatment plan for an individual with ASD (Klinger et al., 2020). Specifically, studies should report more specificity regarding the circumstances of each session related to the structure of the session itself (physical space, activities, goals), the interventionists (credentials and training), and the animals (ages, breeds, training, experience in AAI). Understanding which components drive optimal outcomes will enable practitioners to tailor and enhance AAIs for individuals.

The credentials of the interventionists were reported in 34 out of 36 studies (94%), which demonstrates an important (and significant) increase in reporting since earlier reviews citing only 55% (O’Haire, 2017). Rigorous certifications are important for growth in the field, they scaffold interventionists to facilitate a safe, ethical, and productive interaction between clients and the animals (Fine et al., 2015). To do so, they need to be knowledgeable about the client population they are serving as well as the species of animal and the interactions between the two. Practitioner training should include professional development to uphold high standards and increase the credibility of the field (Fine & Andersen, 2021).

Horses were the most researched animal for AAI and ASD. Further development of the appropriateness of the use of equines in AAI will be important as suggested in another recent equine-focused review (Peters & Wood, 2017). None of the equine studies mentioned specific selection or training programs for the horses involved. In terms of training for canines, 10 studies mentioned an organization of certification or training and five additional studies mentioned that the dogs were trained but did not specify further. This finding is critical to draw attention to the need for appropriate selection of animals for AAI (Fine et al., 2019) and for development of specific protocols that uphold the animal’s welfare (Vitztum & Urbanik, 2016). Training may help to prepare the animal for the interaction and may influence welfare and/or the efficacy of the animal-assisted intervention (Levinson, 1983). Future research is needed to fully understand the role of the animal and the influence of their training on the outcomes of AAI for both the animal and the human involved.

The total contact time reported on average was 15.7 hours. However, after removing outliers with significantly higher contact times (with Cerino et al., 2016 logging 80 hours of equine-assisted intervention and Byström et al., 2019 logging 100 hours of equine-assisted intervention), the average total contact time was 9.23 hours. Future studies should attempt to determine the longitudinal efficacy of AAI by measuring changes post intervention to determine the dose needed to maintain effectiveness demonstrated immediately after the intervention.

Evaluating the Evidence

Though the total number of studies has significantly increased in the last four years, there has not been a comparable increase in rigor of study design. The methodological rigor of individual studies on average met 76% of the methodological considerations assessed (range: 33–100%, Appendix 1). In a recent systematic review of assistance dogs, 62% of studies met the methodological considerations discussed (Rodriguez et al., 2020). The number of studies with sample sizes <20 individuals has remained consistent. With respect to bias, only 47% of studies have a control condition and 9% incorporated blinded raters. Though institutional review board approval was noted for most studies, approval or waiver from an institutional animal care and use committee was only reported for 11% of studies. To focus on welfare of animals in AAI (Fine et al., 2019), future studies should ensure approval or a waiver from the institutional animal care and use committee.

To address improved methodological rigor, future studies should consider larger sample sizes with a control condition, blinded raters, and additional physiological measures to decrease bias. Designs should be selected based on power analyses for appropriate samples sizes as multiple studies in the current review noted issues with power in their limitations. Reporting effect sizes enables comparison across studies and may uncover larger trends in AAI. Recently in the field of human interaction there has been a call to eliminate the “file drawer” effect as there is a preconceived notion that animals are inherently beneficial for humans. Nearly all quantitative studies included in this review published both positive and null findings, indicating important growth in transparent reporting. This review included studies of families (parents) and therapists providing their perspectives and experiences with AAI for ASD. Considering the influence that AAI may have beyond the individual with ASD, especially in the context of service animals that interact with families as well, may uncover additional benefits and challenges of the intervention.

Outcomes of AAI for Autism

To date, no studies have reported significant harm or decline in any area of functioning. Additionally, all negative outcomes of AAI for ASD emerged from qualitative findings and primarily focus upon increased stress, access to the interventions, and the structure of the intervention itself.

Social interaction remains as the most identified outcome in 21 studies both qualitatively (n=7) and quantitatively (n=14). Given the continued research focus on this outcome, it is still the case that social interaction is the most promising outcome due to the number of independent research teams evaluating it and the many ways of analysis. A prominent theory in human-animal interaction literature is social support theory (Beetz, 2017) in that animals can be a nonjudgmental companion and support figure for individuals. Additionally, the theory suggest that animals can function as a bridge for social interactions in the human-to-human context (McNicholas & Collis, 2000). This finding supports this theory in that animals may be a social support for individuals with ASD, creating an avenue for growth in social interaction skills and experiences.

Language and communication were the next most explored outcomes in this review with 14 studies citing related outcomes. Findings were predominantly positive with only two studies not citing significant improvements. Animals may provide an avenue for individuals with ASD to improve communication and language by acting as a bridge for the skills developed during interaction with the animal to interaction with humans (McNicholas & Collis, 2000). Future studies should continue exploring outcomes related to language and communication. Incorporating theory from the field of communication in the context of AAI and ASD may provide structure for explaining the mechanisms that occur within the intervention.

Problem behavior outcomes were reported in about 25% of the studies in this review (n=11) and out of eight different standardized survey measures of problem behaviors, only one measure did not have fully positive results. Improvements to positive emotions were found in four studies out of six studies that explored the outcome. All the studies that reported on stress (n=9) found that AAI can relieve stress for individuals with ASD. Few studies reported on motor skill improvements and restricted and repetitive behavior outcomes. As AAI for ASD is further refined, targeted exploration of these less frequent outcomes will be important before they are categorized as not a significant outcome in AAI for ASD. The current methodology of AAI may not be focused on these outcomes and thus may be influencing their apparent role in the intervention.

Findings from parent, family and therapist focused studies were mixed, identifying multiple benefits and multiple challenges. This review demonstrated an increased focus on outcomes of AAI for ASD beyond the individual with ASD. The increase in quantity of studies focused beyond the individual with ASD may indicate a pivot in the literature to obtain a more complete picture of the role of AAI for ASD incorporating effects on caregivers and family members. Also, this review identified a growing focus on the intervention process itself. Nine studies commented on the process of animal-assisted interventions with 78% focusing on the importance and role of the animal within the session. This new focus is promising as conversations regarding set guidelines and protocols for the field move to the forefront of the literature. Defining the intervention and the process by which it is examined is critical for the development of the intervention as well as recognition for the field as an evidence-based practice for individuals with ASD.

Limitations

Findings from this systematic literature review should be interpreted in the context of several limitations. First, inclusion criteria for this review excluded articles that were not published in English. Second, though many studies did report nonsignificant findings, there may be a publication bias present in that all studies reported at least one significant finding. Additionally, literature reviews/meta-analyses, dissertations, validation of measures, book reviews, book chapters, magazine articles, commentaries, editorials without empirical outcomes or conference presentations, abstracts or posters were not included. Unpublished studies (e.g. dissertations) may show null/negative findings and thus could be important to include in future reviews.

Conclusion

Based on the results of the current review, AAI should continue to be considered as a “possibly efficacious” adjunct intervention for individuals with ASD as proposed by O’Haire (2017). Though the number of studies has substantially increased, findings are still varied due to the many different models of the intervention. Until there are standardized intervention designs and multiple studies completed on those designs, AAI will not be considered an evidence-based practice for ASD. Future studies should focus on first refining the structure of animal-assisted intervention and subsequently develop a protocol to assess those designs, the mediators, and moderators associated. It is important that future studies lead to understanding of the specificity and variability of treatment outcomes. In doing so, researchers need to prioritize the well-being of human subjects as well as prioritize the welfare needs of the animals involved.