Author Name: Natasha Dagger; BSc (Hons) Veterinary Nursing Science
Animal Assisted Therapy (AAT) is an intervention which utilises therapy animals to meet measurable goals and to help overcome physical and mental disorders. Many studies have examined the effectiveness of AAT and human-animal interactions on various aspects of physical and mental health. This research is valuable as it can be shared within the field and utilised to help design successful AAT programmes for the future. There are contradictions between the various studies and the different methods used, although a common justification for the research is that there is a shortage of reliable studies with controlled conditions. Within this research, larger sample sizes are important for allowing the study of statistically significant differences in data. In addition, by ensuring that there is equality between intervention and control groups, the most reliable results will be obtained. Clear and controlled experimental designs, with a detailed range of AAT activities, are necessary in order to allow repeatability for future studies. The overall message of the research is that the benefits of AAT greatly outweigh the risks. The previous studies provide initial evidence in proving the importance of this research, however improved validity and replicability would allow the production of guidelines for AAT procedures, allowing greater adoption of the same protocols. Future research into specific targeted health conditions and their response to AAT would also be beneficial.
Animal Assisted Therapy (AAT) is an intervention which utilises therapy animals to meet measurable goals. It is carried out by trained health professionals and animals and is used on individuals or groups of patients, to help overcome physical and mental disorders (Maber-Aleksandrowicz, Avent and Hassiotis, 2016).
Many studies have examined the effectiveness of AAT and human-animal interactions on various aspects of physical and mental health since 1792 (Bernabei et al., 2013). This research has aimed to provide clearer evidence for the benefits of this practice and to define future research needs (Bert et al., 2016).
2.0 Critical Review
2.1 Authorship and summaries
Braun et al., (2009) set out to determine whether there were significant differences in vital signs and pain relief in children after AAT. They presented an accurate summary detailing the use of immunocompetent children in an acute paediatric care hospital. The importance of the research in terms of its potential for sharing with other researchers which could help with successful programmes for the future is clearly stated (Walters Esteves and Stokes, 2008). In Calcaterra et al., (2015) they aimed to understand children’s pain and stress response in the immediate post-surgical period after AAT. The purpose of the study was stated and the outcomes were defined. Stefanini et al., (2015) had a clear abstract that included objectives for the study plus a thorough design of how they aim to compare the responses of children and adolescents with mental disorders after AAT. In their later study, Stefanini et al., (2016) determined responses of children and adolescents with emotional and behavioural disorders after AAT. This study presented an accurate summary with detailed results (Zirkel, Garcia and Murphy, 2015). The second article by Stefanini et al., (2016) also stated that the objectives were met and that additional results were found, unlike Stefanini et al., (2015) who do not clearly state in their conclusion whether the objectives were met. Only a brief summary is given for Braun et al., (2009) and they state that AAT could have an influence on children and pain but they do not conclude how. Stefanini et al., (2016) used their previous research (Stefanini et al., (2015) as a conceptual framework for their research study (Zirkel, Garcia and Murphy, 2015).
The authors of Braun et al., (2009) all come from a nursing background but have no animal related qualifications. There are also minimal results indicating other articles written by them (Bert et al., 2016). Calcaterra et al., (2015) have two authors who are qualified veterinary surgeons, with experience within their field and who have previously been involved with other studies relating to dogs and pet behaviour. The authors of Calcaterra et al., (2015) have qualifications in child neurology and behaviour which suggests their epistemology of research is justified and supported by their knowledge (Bert et al., 2016). Stefanini et al., (2015) and Stefanini et al., (2016) have three of the same authors involved in their studies. All are experienced within psychology, but like Braun et al., (2009), none of the authors have any qualifications within the animal therapy industry (Bernabei et al., 2013).
2.2 Sampling Methods
Braun et al., (2009) conducted research using participants aged three to seventeen years, paralleling that of Calcaterra et al., (2015) who also used participants of the same age range. In comparison, Stefanini et al., (2015) and Stefanini et al., (2016) used participants aged eleven to seventeen years of age, which does not cover as broad a range as the former studies. However, by targeting secondary school children, they have chosen a specific sample of a developmental age when hormones and behaviour begin to change, which supports their studies of mental disorders (Stefanini et al., 2015). A sample size of ninety-four subjects were used by Braun et al., (2009) which is a large sample size compared to other studies. In contrast, Calcaterra et al., (2015) and Stefanini et al., (2016) both used limited sample sizes of forty participants. Larger sample sizes can be important for indicating statistical differences in data (Gentles et al., 2016). Braun et al., (2009), Calcaterra et al., (2015), Stefanini et al., (2015) and Stefanini et al., (2016) all conducted their experiments with equal intervention and control group sizes. However, it is not clear in both Braun et al., (2009) and Calcaterra et al., (2015) whether the groups were balanced for gender. Comparatively, Stefanini et al., (2015) and Stefanini et al., (2016) provided information on gender and kept their intervention and control groups equal which will produce fairer results (Gentles et al., 2016).
Stefanini et al., (2015) only used participants who were able to give informed consent. By providing their own consent, and being fully informed on the study, participants may be more likely to want this study to succeed which in turn could cause bias in the results (Spiegelman, Rivera-Rodriguez and Haneuse, 2016). The target sample range in Braun et al., (2009) was determined based on the Wong-Baker FACES pain assessment scale. Participants were only included if they were able to use the assessment scale and indicated a pain of two out of ten or higher. This non-probability sampling plan is biased in favour of children who can use the scale (Setia, 2016).
A quasi-experimental intervention design is conducted in Braun et al., (2009) as the participants were in the intervention or control groups dependent on their placement by the researcher. This can cause bias as this method allows the researcher to dictate the groups that participants are placed into; therefore there is a lack of randomisation and realistic results (D’Onofrio et al., 2013). The strength of using this method is that it can make the process quicker as, if the researcher is aware of a specific area from past research, then they can design their groups around this, though the results may be more down to choice rather than chance (D’Onofrio et al., 2013). In comparison, Calcaterra et al., (2015) used a probability sampling method of permuted blocks randomisation; participants picked an envelope each and then they were placed into the group that was on the paper inside. This randomisation can be supported by Stefanini et al., (2015) and Stefanini et al., (2016) whose studies both assigned participants to groups through generating computerised random numbers. These latter methods of producing sample populations prevent selection bias and ensure that each participant has an equal chance of being in the intervention or control group, so are therefore more reliable (Kahan, Rehal and Cro, 2015).
There was a sampling inconsistency in Stefanini et al., (2015) due to the difference in the percentage of the participants who had previously received treatment for psychopathology (26.5%) versus the participants who had not received previous treatment (73.5%). The external validity of this was not well controlled therefore potentially producing treatment bias and inconsistency (Berner et al., 2014). Calcaterra et al., (2015), Stefanini et al., (2015) and Stefanini et al., (2016) are all limited by their small sample size whilst Braun et al., (2009) lack true randomisation.
2.3 Research Approach
Braun et al., (2009) justified their qualitative research into the impact of animals on the alleviation of physical pain in children as no studies had yet been done. This study follows an inductive approach, aiming to produce a new theory validated by data. The research questions were clear and they included a two-tailed hypothesis predicting both relevant directions. Calcaterra et al., (2015) justified their reason for mixed methods research in the same manner by confirming that there had been no studies in this area previously carried out. Stefanini et al., (2015) discuss how most studies have been quasi-experimental or non-experimental and have produced inconclusive findings for animal intervention and mental illnesses. The justification from Stefanini et al., (2016) supports this by adding that there is a shortage of reliable studies with controlled conditions. Stefanini et al., (2016) takes a deductive approach to their research by testing their previous theory outlined in Stefanini et al., (2015) as they already know from this earlier study that AAT was effective.
2.4 Experimental Design
Braun et al., (2009) begin the study by taking baseline measurements of the participants’ blood pressure, respiratory rate and pain level. Post-test measurements were taken to discover differences occurring from pre to post AAT intervention. Calcaterra et al., (2015) also measured baseline blood pressure, respiratory rate and pain level but in addition measured heart rate, oxygen saturation and salivary cortisol levels. Similarly, they carried out post-test measurements but also performed a mid-test measurement. The data in Braun et al., (2009) and Calcaterra et al., (2015) were analysed using a two-tailed test to estimate pain level in AAT. This statistical test suggests that the researchers are conservative in their ideas about the outcome (McHugh, 2013). Comparatively, Calcaterra et al., (2015) used a Fisher exact test for the two nominal variables which is an accurate test when used with a small sample size. However, they may not have considered their small sample size of forty for the two-tailed test, as it is recommended for use in a sample size above fifty. Stefanini et al., (2015) devised structured weekly sessions of forty-five minutes of AAT over a three month period. They performed pre and post-tests of mood and eating disorders. Whilst Stefanini et al., (2016) performed their study over a 14 month period with forty-five minute AAT sessions, testing mood, eating disorders and anxiety disorders. Similarly, Stefanini et al., (2015) and Stefanini et al., (2016) both used a T-test for their independent samples to investigate the differences between behavioural variables with AAT. This test details the difference between the averages in the two groups (Ashbeck and Bell, 2016). Stefanini et al., (2015) used an assessment scale for investigators to measure global functioning in their participants. This scale is suitable for ages six to seventeen years which supports their sample population age. In contrast, both Braun et al., (2009) and Calcaterra et al., measured cardiovascular response and used a Wong-Baker FACES pain assessment scale which is designed for children to self-report their pain level (Koldas Dogan, 2012). In the latter study, Calcaterra et al., (2015) also measured neurological impact plus salivary cortisol levels, which enhances the scientific validity of the data (Koldas Dogan, 2012).
Braun et al., (2009) designed their study to include a fifteen to twenty minute session of AAT with the intervention group, but only a fifteen minute session with the control group. In comparison, Calcaterra et al., (2015) ensured that the intervention and control group sessions were equal at twenty minutes each. Additionally, Braun et al., (2009) did not specify how long it was after the dog and handler had left the room that they took post-test measurements, therefore this reduces repeatability. They also did not state what breed of dog that they used for their AAT unlike Calcaterra et al., (2015) who explained the breed and age of dog used. Another factor that reduces repeatability in Braun et al., (2009) concerns pharmacologic pain relief. The participants in their study were not denied analgesia pre, during or post intervention. Whilst ethically sound this means that each participant could have been receiving differing types of analgesia at various times throughout the study which could influence the participant’s response. Stefanini et al., (2016) varied their sessions between a hospital room and a garden dependent on weather, which may cause unreliability in results. Comparatively, Calcaterra et al., (2015) avoided confounding variables by providing all participants with the same dose of analgesia twenty minutes before the end of surgery. Stefanini et al., (2015) used video tapes to observe their AAT sessions, allowing them to use this to view the sessions as many times as they felt necessary, ensuring important details were not missed. This can also limit the Hawthorne effect from influencing the behaviour of the participants (MacNeill et al., 2016). They had a clear experimental design with a range of AAT activities including play, grooming, cleaning, agility, obedience and walking (Stefanini et al., 2015). The authors also provided the reader with details of these activities, unlike the other researchers where it is unknown what AAT activities were carried out in the study sessions.
2.5 Significance of Findings
The results of Braun et al., (2009) show that 12.3% of the participants had previous AAT experience and 63.2% had a pet of their own, which may be a potential source of bias. There was no significant statistical difference between the intervention and control results which were very similar. They did not reach their intended sample size due to the death of the therapy dog, but chose to publish the results they had obtained instead of finding another dog. Although this means that they have strong internal validity by testing the causal relationship, it lacks the potential for comparison of results. For example, Stefanini et al., (2015) used several therapy animals and found that there were significant differences between the intervention and control results, but due to extraneous variables, this may be unreliable. Braun et al., (2009) found an average time difference of 107 minutes between analgesia administration of the intervention and control group which may have an effect on the results and is a potential flaw of the study. Stefanini et al., (2016) report that they were able to accept their hypothesis as their intervention group had significant improved mental wellbeing.
2.6 Addressing Aims and Future Research
Braun et al., (2009) finished their study early, due to the therapy dog’s death and although they did not reach their desired sample population, they still managed to reject their null hypothesis by producing impartial results. Similarly, Calcaterra et al., (2015) also met their aims in this pilot study but further research with more participants is needed for both studies, due to their limited sample sizes. Stefanini et al., (2015)’s study faced several limitations. Again the sample size was small compared to the population, but also, they lacked realistic comparisons for the participants’ behavioural change. Therefore, further follow up research is needed. Stefanini et al., (2016) met their aims in their study with promising results, however, further research is needed using randomised controlled studies. It would be useful to explore and define which mental health conditions and physical pain impairments respond to AAT most effectively, by utilising more specific target populations (Stefanini et al., (2015). These studies of AAT effectiveness provide initial evidence, yet, there is a clear critical gap in research on published protocols for guidelines on AAT procedures, and it would be helpful if researchers designed a set methodology that all hospitals could follow (Stefanini et al., (2016).
How AAT impacts patients with physical pain and mental health conditions has been widely researched. Studies confirm that AAT can be an effective therapeutic treatment to alleviate pain in immunocompetent children, after surgery and also can result in positive changes in mental health. The overall message of the research is that the benefits of AAT greatly outweigh the risks.
The work to date provides initial evidence of the importance of this research in creating successful AAT programmes, but there is still potential for greater validity and the production of guidelines for AAT procedures. This would allow hospitals to adopt the same protocol. Future research into specific targeted health conditions and their response to AAT would also be beneficial.
Ashbeck, E. L and Bell, M. L. (2016) Single Time Point Comparisons in longitudinal randomized controlled trials: power and bias in the presence of missing data. Bmc Medical Research Methodology [online]. 43 (16), pp. 1-8. [Accessed 19 October 2016].
Bernabei, V., De Ronchi, D., La Ferla, T., Moretti, F., Tonelli, L., Ferrari, B. and Forlani, M. (2013) Animal-assisted Interventions for Elderly Patients Affected by Dementia or Psychiatric Disorders: A Review. Journal of Psychiatric Research [online]. 47 (6), pp. 762-773. [Accessed 18 October 2016].
Berner, J., Fossell, R., Ha, S Y, Hacker, J.P. and Snyder, C. (2015) Increasing the Skill of Probabilistic Forecasts: Understanding Performance Improvements from Model-Error Representations. 2015 American Meteorological Society [online]. 143, pp. 1295-1320. [Accessed 20 October 2016].
Bert, F., Gualano, M.R., Camussi, E., Pieve, G., Voglino, G. and Siliquini, R. (2016) Animal Assisted Intervention: A Systematic Review of Benefits and Risks. European Journal of Integrative Medicine [online]. 8 (5), pp. 695-706. [Accessed 10 October 2016].
Braun, C., Stangler, T., Narveson, J. and Pettingell, S. (2009) Animal-assisted Therapy as a Pain Relief Intervention for Children. Complementary Therapies in Clinical Practice [online]. 15 (2), pp. 105-109. [Accessed 10 October 2016].
Calcaterra, V., Veggiotti, P., Palestrini, C., De Giorgis, V., Raschetti, R., Tumminelli, M., Mencherini, S., Papotti, F., Klersy, C., Albertini, R., Ostuni, S. and Pelizzo, G. (2015) Post-operative Benefits of Animal-assisted Therapy in Pediatric Surgery: A Randomised Study. Plos One [online]. 10 (6), pp. 1-13. [Accessed 10 October 2016].
D’onofrio, B, Lahey, B.B., Turkheimer, E., Lichtenstein, P. (2013) Critical Need for Family-based, Quasi-experimental Designs in Integrating Genetic and Social Science Research. American Journal of Public Health [online]. 103 (1), pp. 1-11. [Accessed 20 October 2016].
Gentles, S., Charles, C., Nicholas, D., Ploeg, J. and McKibbon, K.A. (2016) Reviewing the Research Methods Literature: principles and strategies illustrated by a systematic overview of sampling in qualitative research. Systematic Reviews [online]. 172 (5), pp. 1-11. [Accessed 19 October 2016].
Kahan, B., Reha, S., Cro, S. (2015) Risk of Selection Bias in Randomised Trials. Trials [online]. 405 (16), pp. 1-7. [Accessed 18 October 2016].
Koldas Dogan, S, Ay, S., Evcik, D., Kurtais, Y. and ÖZtuna, D.G. (2012) The Utility of Faces Pain Scale in a Chronic Musculoskeletal Pain Model. Pain Medicine [online]. 13, pp. 125-130. [Accessed 12 October 2016].
Maber-aleksandrowicz, S., Avent, C. and Hassiotis, A. (2016) A Systematic Review of Animal-assisted Therapy on Psychosocial Outcomes in People with Intellectual Disability. Research in Developmental Disabilities [online]. 49-50, pp. 322-338. [Accessed 18 October 2016].
Macneill, V., Foley, M., Quirk, A. and McCambridge, J. (2016) Shedding Light on Research Participation effects in behaviour change trials: a qualitative study examining research participant experiences. BMC Public Health [online]. 91 (16), pp. 1-8. [Accessed 18 October 2016].
McHugh, M.L (2013) The Chi-square Test of Independence. Biochemia Medica [online]. 23 (2), pp. 143-149. [Accessed 20 October 2016].
Setia, M. S (2016) Methodology Series Module 5: Sampling Strategies. Indian J Dermatol [online]. 61 (5), pp. 505-509. [Accessed 20 October 2016].
Spiegelman, D., Rivera-Rodriguez, C.L. and Haneuse, S. (2016) Evaluating Public Health Interventions: 3 The Two-Stage Design for Confounding Bias Reduction—Having Your Cake and Eating It Two. Am J Public Health [online]. 106 (7), pp. 1223-1226. [Accessed 18 October 2016].
Stefanini, M.C., Martino, A., Allori, P., Galeotti, F. and Tani, F. (2015) The Use of Animal-assisted Therapy in Adolescents with Acute Mental Disorders: A Randomised Controlled Study. Complimentary Therapies in Clinical Practice [online]. 21 (1), pp. 42-46. [Accessed 12 October 2016].
Stefanini, M.C., Martino, A., Bacci, B. and Tani, F. (2016) The Effect of Animal-assisted Therapy on Emotional and Behavioural Symptoms in Children and Adolescents Hospitalised for Acute Mental Disorders. European Journal of Integrative Medicine [online]. 8 (2), pp. 81-88. [Accessed 12 October 2016].
Walters Esteves, S and Stokes, T. (2008) Social Effects of a Dog’s Presence on Children with Disabilities. Anthrozoos [online]. 21 (1), pp. 5-15. [Accessed 10 November 2016].
Zirkel, S., Garcia, J.A. and Murphy, M.C. (2015) Experience-sampling Research Methods and Their Potential for Education Research. Educational Researcher [online]. 44 (1), pp. 7-16. [Accessed 10 November 2016].