Animal Assisted Therapy

Early institutional uses of animals

Historians and reviewers trace the therapeutic use of animals to the York Retreat, a Quaker institution for people with mental illness in England. A literary-historical study examines the uses of non-human animals during the Retreat's early years, 1796 to 1813, and argues that the figure of the animal came to represent the patient's docility rather than unreason. A review by Fine, Beck and Ng instead dates the first recorded use to 1792 and mentions rabbits and poultry, so sources differ on the date.

The same review says Florence Nightingale advocated animal companionship for patients, and gives the year as 1869. That date was not checked against her writings for this draft. These accounts concern institutional and nursing practice and do not describe a method resembling modern structured therapy.

Levinson and research in the twentieth century

According to Fine and colleagues, the child psychologist Boris Levinson observed in the 1950s and 1960s that a withdrawn child engaged with his dog, Jingles, and in 1964 he coined the term pet therapy. The review places empirical research by Samuel and Elizabeth Corson at Ohio State University in the 1970s, and credits them with the idea of animals as a social lubricant. These attributions come from one secondary review and primary publications were not opened.

The same authors state that early work relied heavily on anecdote and on non-validated instruments, that media emphasis on positive outcomes created a confirmation bias, and that one author warned against presenting animal-assisted work as a panacea. They describe the field as now moving towards mixed-methods research, physiological markers and meta-analyses.

Intervention, therapy, activity and education

The International Association of Human-Animal Interaction Organizations (IAHAIO) published a white paper in 2014 on definitions for animal-assisted intervention and on the wellness of the animals involved, and updated it in 2018. An umbrella review summarises its categories. Animal-assisted intervention (AAI) is a goal-oriented, structured intervention that intentionally includes animals in health, education and human services. Animal-assisted therapy (AAT) is a goal-oriented, planned, structured intervention directed or delivered by health, education or human service professionals. Animal-assisted activity (AAA) covers informal visits and interactions, and animal-assisted education (AAE) is delivered by qualified education professionals.

Inconsistent use of the terms

Definitions are not used uniformly. One review of therapy dog welfare describes AAT as requiring documentation and evaluation and AAA as spontaneous and lacking a defined goal, while a youth mental health review treats AAT as a subcategory of AAI. The umbrella review of 15 systematic reviews found terminology frequently misapplied, with studies labelled as AAT lasting from four consecutive days to 18 months and sessions from six minutes to full days. The authors called for globally recognised terminology.

Equine programmes use their own labels. A meta-analysis on autism defines equine-assisted activities and therapies (EAAT) as programmes that incorporate horses for rehabilitative and educational benefit, and separates equine-assisted therapies, which include hippotherapy and equine-assisted psychotherapy, from equine-assisted activities such as therapeutic horseback riding. Pet therapy is the older label attributed to Levinson.

People and animals involved

A session involves a person receiving the intervention, a trained animal and usually a handler, working under a professional who directs the therapy. The umbrella review found that operators in the included studies were dog or animal handlers, psychologists and therapists, and that training specifications were inconsistent. Dogs were the main species in all of those reviews but one, and horses were the second most common.

A review of youth studies from 2000 to 2015 found 11 of 24 studies used horses and 10 used dogs, with the rest using other animals such as guinea pigs. Most were delivered in outpatient settings such as stables, classrooms, advocacy centres and medical or dental offices, with fewer in residential treatment and one in an inpatient psychiatric unit.

Variation in sessions

No single protocol exists. The youth review noted that few studies used a manual and that none had examined mechanisms of change. A review of therapy dog welfare reports suggested limits of about 60 minutes for a session and about three sessions a week for a dog, but these are suggestions drawn from small studies and not a standard.

A hospital-infection review advises an assessment of the person before animal contact is started, with the goals of the therapy defined, and Fine and colleagues describe a behavioural and veterinary evaluation of the animal beforehand. The sources reviewed do not describe a standard intake procedure.

Populations and purposes reported

In the umbrella review, 60% of the reviews concerned mental health disorders, and others concerned dementia and autism spectrum disorder. A 2026 meta-analysis of randomised trials in adults examined depression, anxiety, stress, pain and gait, and a review in healthcare settings examined pain relief. Equine programmes have been studied for children with autism, and dog programmes for children affected by trauma.

Other species and uses

A review of 135 articles on less common species covered guinea pigs, rabbits, farm animals, alpacas, donkeys, reptiles, aquarium fish and dolphins. Most reported positive outcomes for people, but the authors found little information on effects on the animals. For dolphins they note limited evidence of effectiveness and describe them as non-domesticated animals whose captivity limits consent to participate.

Adults: depression, anxiety, stress and pain

A 2026 meta-analysis searched six databases to August 2024 and included 35 randomised trials with 2,391 adults, with or without illness. At the end of the intervention AAT was associated with lower depression (Hedges' g -0.403, 95% confidence interval -0.536 to -0.271), anxiety (-0.661, -1.069 to -0.253) and stress (-1.062, -1.849 to -0.275). The authors reported substantial variability between studies and no meaningful effect on pain or gait, and found that psychological effects depended on population and intervention characteristics.

A systematic review of pain in healthcare settings included 24 studies with 1,950 participants, of which 10 were randomised. Thirteen of the 18 studies reporting pre-to-post change in pain found a significant reduction, but the authors judged the evidence weak because of design and quality problems.

Dementia, schizophrenia and young people

A meta-analysis of 11 randomised trials with 825 people with dementia (searches to November 2021) found a reduction in behavioural and psychological symptoms, especially depression, but no significant improvement in cognitive function, daily living, agitation or quality of life. A review of seven randomised trials in schizophrenia and related disorders found too much heterogeneity for meta-analysis, with mostly no significant effects on symptoms or quality of life, and judged the evidence inconclusive.

A review of 24 youth studies from 2000 to 2015, 11 of them randomised, rated findings as positive in 8, mixed in 10 and without effect in 6. Equine therapy for autism and canine therapy for childhood trauma looked most promising, but the authors described the research base as slim.

Equine-assisted programmes for autism

A 2023 systematic review and meta-analysis of 25 articles from 2009 to 2022, covering 623 children with autism, reported improvements in social cognition, communication, irritability, hyperactivity and language, but not in social awareness, motivation, lethargy or stereotypy. Eight studies were randomised, and the authors noted that blinding of participants and assessors was difficult. Evidence on motor and sensory function and on family functioning was insufficient.

Limits of the evidence

Across these reviews the interventions, species, doses and comparators differ widely, and participants cannot be blinded to the presence of an animal. The umbrella review concluded that heterogeneity was widespread, and the youth review found no studies of mechanisms. Oxytocin and cortisol are being measured as possible markers, but a proposed mechanism is not evidence of benefit. Findings should not be transferred between populations, species or outcomes.

Welfare of therapy dogs

A review of therapy dog welfare identified possible stressors, including inappropriate physical contact, novel environments, high room temperature, limited water, contact with children under 12, too little rest and a lack of freedom to withdraw. Across nine studies of cortisol the results were mixed, and the author concluded that clear conclusions on the effect on dogs' wellbeing were lacking because of small samples and heterogeneity.

Wider welfare and consent questions

Fine and colleagues describe a cost and benefit balance across the animal's working life, including behavioural and veterinary evaluation, matching the animal's temperament to the work, and planning for retirement. The review of unconventional species found that effects on the animals, especially farm animals and reptiles, were largely overlooked, and that reliable stress indicators were lacking for some species.

Infection, bites and allergy

A 2000 review in the Journal of Hospital Infection named animal bites, allergies and zoonoses (infections passed from animals to people) as the major animal-associated hazards in healthcare settings. It advised a full assessment before starting, defined goals, and policies covering the suitability of patients and animals and infection control.

A university programme study screened 22 dogs and 2 cats in 118 tests. Seventeen were positive, 14 for potentially zoonotic organisms, nearly half of the positives being Campylobacter, and none was positive for MRSA or MRSP. Of 40 handlers surveyed, 70% had no concerns about transmission and adherence to infection control was lacking. The study was small, relied on self-report and covered one university, so it does not measure the rate of infection in patients.

Suitability, training and regulation

Fine and colleagues state that handler training should cover zoonotic disease, infection control, what to do after an accident or injury, and reading animal body language, and that no universal standards exist for continuing education and certification. The umbrella review likewise found no clear regulations on the training of operators. IAHAIO has published guidelines on minimum education and training standards. No statutory regulation or training standard was found in the sources reviewed.

None of the reviews examined provides a list of formal contraindications, and adverse events in people were not systematically reported in the sources used. Animal-assisted therapy is not a substitute for assessment or treatment of a mental or physical health condition, and care for such a condition should not be delayed or replaced.