Effects of Dog-Based Animal-Assisted Interventions in Prison Population: A Systematic Review

Simple Summary Inmates usually suffer from depression, anxiety, or loss of empathy. Thus, animal-assisted therapies, and in concrete dog-based assisted therapy, were introduced in prisons in order to ameliorate the consequences of being into prison. Therefore, the aim of this study was to provide an up-to-date analysis of the research on the effects of dog-based animal-assisted therapy in the prison population. Results showed that mental health, emotional control, empathy, or academic skills could be improved after dog-based assisted therapy. Most of the studies included activities related to dog training, dog caring, or activities. The duration of dog-based therapies ranged between 60 and 120 min, with the frequency being between 1 and 3 days/week. In conclusion, these potential enhancements might lead to a reduction in recidivism and violence. Abstract Background: Animal-assisted interventions, in concrete dog-assisted intervention, have been introduced in prisons to reduce recidivism as well as to improve the well-being of prisoners. Therefore, the aim of the present systematic review is to provide an up-to-date analysis of the research on the effects of dog-based animal-assisted therapy in prison population. Methods: An electronic search of the literature was performed, and 20 articles were included. The PRISMA guideline methodology was employed. Results: Included studies involved a total of 1577 participants. The vast majority of protocols included activities related with dog training, dog caring, or activities, which included vocational or educational components. Duration of dog-based therapies ranged between 60 and 120 min, with the frequency being between 1 and 3 days/week. Statistically significant improvements in prisoners were observed in 13 studies. Conclusions: Dog-based animal-assisted therapy may improve anxiety, stress, recidivism, and other social variables in male or female inmates.


Introduction
The proportion of both female and male inmates has been increasing since 2000. In this regard, the worldwide rate of female inmates has increased 50.2%, while the worldwide rate of male inmates grew 18.1%. Moreover, in the United States (EEUU), it is estimated that 221,600 young people were incarcerated [1] because of antisocial behaviors such as, theft, vandalism, bullying, or use of weapons [2]. People involved with criminal justice frequently are exposed to violence, traumatic experiences, or critical life events. Thus, the prevalence of posttraumatic stress disorder among sentenced prisoners

Risk of Bias
Risk-of-bias assessment was summarized in Table 1. The lack of control or comparison group (only 25% of the articles fulfilled this criteria) [32,[44][45][46][47], random assignment of participants to intervention (15% of the articles fulfilled this criteria) [44][45][46], random selection of participants for assessment (0% of the articles fulfilled this criteria), or the follow-up rate (25% of the articles fulfilled this criteria) [39,[45][46][47][48] are the most critical concerns. Table 2 shows the population characteristics for each article. A total of 1295 participants are included in the present systematic review comprising ages between 13 and 69.

Study Design
Moreover, three articles (15% of the total) were classified as non-randomized controlled trials [32,47,51] (two groups with pre and post assessments but without randomization) and three (15% of the total) were classified as randomized controlled trials [45,46,50]. The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.

Collica-Cox (2018)
Animals 2019, 9, x 5 of 18  The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable. The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.     The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable. 5   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.         The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.

Jasperson (2015)
Animals 2019, 9, x 5 of 18       The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.       The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.

Koda (2015)
Animals 2019, 9, x 5 of 18     The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.     (2017) Cooke (2015) Dell ( The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   (2017) Cooke (2015) Dell (    The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.     The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.        The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.           The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.         The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.   The Evidence Project risk-of-bias tool rated as: yes, no, not applicable, or not reported. Green: yes; red: no; orange: not reported; white: not applicable.

Minton (2015)
5 of 18 f randomized and non-randomized controlled trial using the Evidence Project risk-of-bias tool.    Table 1. Risk of bias of randomized and non-randomized controlled trial using the Evidence Project risk-of-bias tool.                                    Table 2 shows the intervention protocols in the treatment group for each article. In all the studies, the protocol includes a group of inmates who performed an AAI. Among the activities included in these programs were dog walking, dog training, taking care of dogs or rescue dogs as well as using the dog as emotional support during the therapy. Further details of intervention are summarized in Table 3. Relationships; and CPR, First Aid, and AED certification for adults, children, and infants. The dogs will serve as emotional support during the class when difficult topics are discussed and the dogs will be incorporated into each lesson and serve as avatars/surrogates as women practice some of their skills. The therapy dogs will be available for the children and family members during the reunification/graduation day. Experimental group is involved in dog-assisted therapy session, while control group is only part of standard rehabilitation program. Participants had to experience the interaction with the dog and they were involved in management and performance activities. The treatment group implemented the use of a dog in order to facilitate the educational and therapeutic goals. In general, sessions were focused on the development of social skills, coping skills, and self-awareness. Each week, treatment group would sit in a circle on the floor and the dog would remain in the center of the circle. So, human-animal interaction was based on group member or animal initiative. Maximum 10 young men participating in each session. Participants learns how to train and care for the dogs, so they design training plans and use positive reinforcement methods to achieve their training goals.

Minke (2017)
---Participants are involved in activities such as walking dog, cooking and dining, manufacturing key-hangers, and engaging in hobbies.

Minton (2015)
26-208 --Dog training program in which women participated as dog trainers or assistants for the prison pup program.  Dog-assisted group therapy is involved in competence and communication training that aims to enhance the social and emotional skills of the participants learning through interaction with the dog based on social and emotional skills that humans can learn from canines or socio-emotional interactions.

Syzmanski (2018)
10 120 2 Experimental group was learning to train dogs, while control group was walking the dogs. They also had classroom-based didactic sessions each week that focused on information about dog care, dog behavior, and humane treatment. Table 2 shows the control group protocols for each article. Fourteen of them [14,18,39,44,48,49,[52][53][54][55][56][57][58][59] did not include a control group, while the design of remaining six studies [32,[45][46][47]50,51] did. In this regard, treatment as usual or activities without dog presence were the most usual protocols for control groups. Only one article of those which present a control group included a passive control group (where participants did not participate in any therapy or activity) [32]. Further details can be checked in Table 2.

Outcomes
In order to summarize the outcomes, Table 4 (for quantitative studies) and Table 5 (for qualitative studies) were created. Table 4 shows all the instruments and outcomes measures for the quantitative studies. Regarding the most studies variables in these studies, anxiety was measured in six articles [32,39,47,49,54,57], depression was measured in four of them [39,47,54,57], and, lastly, tension, irritation, vigor fatigue, distraction, and perspective taking were measured in two articles [32,39]. Three studies were classified as randomized controlled trials [45,46,50], two of them [45,50] showing within group improvements in social role, symptom distress, interpersonal relationships, or empathic concern, among others. Moreover, two articles were classified as non-randomized controlled trials [32,47] (two groups with pre and post assessments but without randomization), showing between groups effects on depression, sleep disorders, infraction rate, or state anxiety. The other seven articles [18,39,48,49,54,56,57] were observational studies, with reporting effects on recidivism, emotional state, tension, cortisol level, or emotional stability among other variables.
Furthermore, Table 5 shows the instruments and outcomes for the qualitative studies. One article was classified as a randomized controlled trial [51] and seven as observational studies [14,44,52,53,55,58,59]. Anxiety, depression, symbolism of the rescue dog, universal support, and psychological and emotional health were the most common outcomes among the qualitative studies. Other outcomes such as stress, self-stem, or empathy were also measured. The studies obtained the data using interview, report, observation, and survey questions (see Table 5 for further details).

Discussion
The aim of the present systematic review was to evaluate the effects of dog-based AAI in prisoners. Most of the included articles support the relevance and usefulness of this kind of therapy for male and female inmates. In this regard, many different variables could be improved after dog-based AAI in prisoners, including mental health variables, such as anxiety or depression, emotional outcomes, and other very relevant variables for prisoners, such as empathy, self-control, and even academic skills. The potential enhancements in those variables could also lead to a reduction in recidivism, which was also suggested in the previous review conducted by Cooke and Farrington [41]. However, although there are 20 articles included in this systematic review, given that some concerns may have increased the risk of bias, the interpretation of results must be done with caution.
The mechanisms that explain the improvements of these interventions are not usually investigated in AAI studies or in the analyzed studies. Furthermore, the complexity of the target population leads to a great heterogeneity of the studies and a poor methodological quality of many of them. This makes it difficult for this systematic review to draw conclusions in relation to the different mechanisms and specific strategies used in AAI.
A total of 1295 prisoners were included in this systematic review. In this regard, a similar number of men and women were included. The vast majority of the therapies were conducted in adults [14,18,32,39,[47][48][49][50][51][52][53][54][55][56][57], in which usually both men and women equally benefited from the therapy. However, one intervention [57] reported evidence that dog-based AAI could be more affective among men compared with women. Authors hypothesized that this could be due to the observed better health status of males and also due to the lower stress factors compared to women. In this regard, it must be noted that female inmates often come from disadvantaged backgrounds and have poor education [57], and the frequency of physical and sexual abuse before going to prison is high [60]. Therefore, potential differences between male and female inmates may be relevant when designing AAI programs, but this hypothesis must be confirmed in future studies.
Regarding interventions duration and frequency, eleven articles [14,18,44,45,47,50,51,53,55,57,59] showed effects on relationships, psychological problems, empathic concern, emotional state, personal skills, and control of emotions. The proposed dog-based AAI in the vast majority of those studies had a duration of 60 min, one day a week. However, previous studies focused on psychological therapies indicated that the longer the intervention the greater the effects [61,62]. Thus, future studies should investigate the role of duration in the effectiveness of the AAI. In terms of the activities to be carried out, the most beneficial are related to keeping in touch with rescued dogs, dog training, and therapies focused on educational and interaction components. Moreover, future dog-based AAI should take into account some of the essential principles proposed by Samhsa [63]. In this regard, dog-based interventions should fulfil some criteria such as (1) safety; (2) trustworthiness and transparency; (3) peer support and mutual self-help; (4) collaboration and mutuality; (5) empowerment, voice, and choice; and (6) cultural, historical, and gender issues.
Regarding the design of the studies, only three of the twenty articles were randomized controlled trials. Therefore, results must be taken with caution, since the risk of bias and quality assessment showed critical issues in the vast majority of the articles. The lack of a comparison group, randomization, or the follow-up rate are the most critical concerns. Furthermore, the large heterogeneity of the people in prison in terms of sex, gender, age, educational background, mental disorders, drug addiction, etc., must be considered. Thus, larger randomized controlled trials are needed, and studies that follow animal-assisted intervention guidelines [64] are encouraged to enable the extraction of strong recommendations and conclusions. Nevertheless, the exceptionality of this research field, the nature of participants as well as situational and environmental context concerns make it almost impossible to achieve these research standards.
There are some limitations that should be considered. First, articles in English, French, Italian, Portuguese, or Spanish were included in the systematic review, so it is possible that studies in other languages were not found. Second, the quality of the articles (only three randomized controlled trials) and the heterogeneity of participants and outcomes mean that the conclusions of the systematic review must be taken with caution.

Conclusions
Dog-based AAI could be a useful tool to improve many different variables including mental health, emotional control, empathy, or academic skills in male and female inmates. However, the methodological quality of the included studies was not optimal, and the heterogeneity of the participants and outcomes was large. Thus, further studies with higher methodological quality are required and subgroups are encouraged to enable the extraction of strong recommendations and conclusions. Funding: The author SV was supported by a grant from the regional department of economy and infrastructure of the Government of Extremadura and the European Social Fund (PD16008).