Schizophrenia and related psychotic disorders are characterized by positive symptoms, negative symptoms, and cognitive difficulties. Hallucinations, delusions, and disorganized speech are examples of positive symptoms, while amotivation, anhedonia, and affective flattening are examples of negative symptoms. Genetic predisposition, substance use, trauma, and acute stress are among the risk factors for development of severe psychotic disorders. In addition, neurobiological factors, such as dopamine dysfunction, are associated with presence of positive symptoms and negative symptoms, as well as cognitive difficulties (1, 2). Overall lifetime prevalence for schizophrenia and related disorders is stated as 7.49 per 1,000 (3). The prognosis varies among individuals and extends between recovery and a chronic, lifelong course (4). Life expectancy is reduced by several years, with somatic comorbidity as one of the major causes (5). Overall, severe psychotic disorders are associated with a high burden of disease (6).

Treatment recommendations consist of a combination of pharmacological and non-pharmacological interventions (7). Current antipsychotic medications are shown to be more effective for positive symptoms than for negative and cognitive symptoms (1), and the latter two symptom groups are important determinants of disability (8). Numerous non-pharmacological interventions are considered in the guidelines by the Norwegian Directorate of Health (7), in accordance with international standards. Psychoeducation, family interventions, cognitive therapy, physical activity, and music therapy are among the included options. However, a substantial group of individuals diagnosed with schizophrenia and related disorders do not experience sufficient symptom relief (9). The heterogenous pathophysiology and phenotypes of severe psychotic disorders underpin the need for varied treatment options (10). Direct interpersonal engagement can be too demanding in some individuals. Interaction with therapeutic animals might theoretically be a less stressful alternative.

Animals have been included in the treatment for several disorders through centuries (11). Currently, there has been a development where anecdotal evidence to a larger extent is replaced by scientific research (12). The International Association of Human-Animal Interaction Organizations (IAHAIO) (13) has published specific guidelines for animal-assisted interventions (AAI). These guidelines are stating that AAI must be targeted and structured, with therapeutic benefits as purpose. Animal-assisted therapy (AAT) and animal-assisted activity (AAA) are two examples of AAI relevant to health care. While AAT must be planned, measurable, and documented, AAA signifies informal interaction. The guidelines are further stating that AAT is targeted toward physical, cognitive, behavioral, and/or socio-emotional functioning, while AAA is targeted toward motivation, education, and/or recreation. Knowledge related to health and behavior of included animals is required for providers of both AAT and AAA. Professional expertise, for example within health care, is in addition required for providers of AAT.

Studies have suggested treatment effects related to AAI for a range of health conditions and diseases (14). Biophilia, stress buffering, and distraction are elements in some theories and hypotheses that may explain potential effects (15). The biophilia hypothesis describes the affinity of humans to other living species (16). Effects related to the biophilia hypothesis may involve feelings of safety and facilitation of interpersonal interactions where animals may serve as social catalysators (17). In addition, decreased levels of cortisol and increased levels of oxytocin, β-endorphin, prolactin, phenyl acetic acid, and dopamine have been detected after interaction with dogs (18). These changes may be associated with physiological and psychosocial benefits, such as stress relief and improvement of social bonding and learning (18–21). Summarized, AAI are aimed at a wide range of symptoms and features, including those presented in severe psychotic disorders. Increased motivation for therapeutic activities due to interaction with animals has been described, for example in a study including individuals with acquired brain injury (22). Treatment effects of AAI will be highly relevant to investigate further for individuals with severe psychotic disorders. This is particularly justified by the fact that lack of motivation, which affects adherence to treatment, is a core feature among the negative symptoms (23).

A systematic review (SR) from 2018 on equine-assisted interventions indicated potential effects for individuals diagnosed with schizophrenia and related disorders. Significant improvement was shown for several outcome measures, such as negative symptoms, social functioning, pharmacological compliance, and risk of violence. The authors stated that further research is needed (24). A SR from 2019, including randomized controlled trials (RCTs) on AAI with several animal species, found inconclusive results regarding treatment effects for individuals diagnosed with schizophrenia and related disorders. However, potential benefits were found for some outcome measures, such as positive symptoms, negative symptoms, emotional symptoms, and self-view (25).

As different animal species have different properties, we sought to investigate effects of dog-assisted interventions (DAI) specifically to increase directness and complement previous SRs. An investigation of therapeutic effects of DAI is also relevant due to findings in a survey among individuals diagnosed with schizophrenia, indicating that the dog was a preferred animal (26). A meta-analysis found that dogs were the most commonly involved animal in AAT (27). Beneficial therapeutic effects may be related to the cognitive and emotional capacities in dogs, in addition to an evolutionary connection with humans (28). Feasibility is also an important issue as dogs can thrive in same environments as humans. We sought to evaluate effects of targeted and structured interventions with therapeutic benefits as purpose. Therefore, both AAT and AAA were included.

Due to an existing knowledge gap, in addition to an extension of the field by four articles published during 2021–2023 (29–32), we found it relevant to perform a modified and updated SR on the topic. Summarized, modifications consisted of broader inclusion regarding study designs, and a narrower approach regarding the objective. The aim of the SR was to investigate effect of DAI for adults diagnosed with shizophrenia and related disorders. To our knowledge, this isolated topic has not been specifically covered by previous SRs.

The SR was designed in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (33). In addition, a document with examples from the guidelines was used (34). Two handbooks, by Cochrane (35) and by the Norwegian Institute of Public Health (NIPH) (36), were also used as references.

In this SR, exclusively including studies with isolated results for adults diagnosed with schizophrenia and related disorders, numerous outcomes of DAI were examined. Both significant improvement and non-significant findings for the intervention groups compared with the control groups were reported for general symptoms, positive symptoms, negative symptoms, anxiety, living skills and quality of life. Significant improvement in the intervention groups compared with the control groups was also described for emotional symptoms, stress, self-esteem, self-determination, social function, communication and interaction skills, lower body strength, and adherence, but each of these outcome measures was only examined in single studies.

Within intervention groups, significant improvement was in addition described for salivary cortisol and social adaptive function, also examined in single studies. One study indicated significant deterioration of non-personal social behavior within the intervention group. Specific investigation of wellbeing, depression, agility, mobility, global cognitive function, alpha-amylase, and extent of social support was performed in single studies, and these outcome measures had no significant changes. Significance for non-verbal communication was not stated.

Heterogeneity in study design and lack of statistical calculations complicated the assessment of study outcomes. Therefore, we considered the findings in relation to factors that may have influenced results. Positive symptom score was the outcome measure with the most convincing findings. Significant improvement was demonstrated in several studies, but it should be noted that this was the outcome measure investigated by most studies. Findings concerning negative symptoms, the second most investigated outcome measure, were more divergent. Overall, inconsistent results were reported for the majority of outcome measures examined by more than one study. In the following sections, we highlight some potential explanations.

Importantly, a substantial difference across the studies was related to the content of the control groups. As an example, the studies by Villalta-Gil et al. (44) and Calvo et al. (39) included specific treatment programs focusing on psychosocial aspects with and without DAT. On the other hand, Chu et al. (45) compared AAA with treatment as usual. While significant improvement for several outcomes occurred within the groups in the first two studies, results from the latter contrasted the two abovementioned studies with a substantially larger degree of significant effects of active treatment. Active intervention also occurred in control groups in other studies—e.g. therapeutic recreation (music, art and education) as comparator in the study by Barker et al. (47). No significant changes for anxiety were seen between the groups in this study. The findings were contrasted by results in the study by Lang et al. (42) where the presence of a dog seemed to be the only difference between the groups (42). The findings suggest that the content in the control group may contribute largely to the heterogeneity of results across studies. As there are many uncertainties related to the effects of components only presented in DAI, a specific recommendation for future research is to conduct component studies. This suggestion is in accordance with a SR regarding factors of AAI (60) and a study on the role of common factors in psychotherapy (61).

A second issue contributing to lack of significant results may be related to other aspects of study design: low numbers of participants or short duration of interventions. As an example, the study by Shih et al. (32) showed significant improvement for quality of life between the groups. Increased overall quality of life was not shown in the other studies, neither between the groups or within the groups. The study by Shih et al. (32) stood out with a higher number of participants. Similarly, in one study showing significant improvement for the intervention group compared with the control group for both positive and negative symptoms, the population consisted of 40 participants (29). For the four studies with non-significant changes for negative symptoms, the samples were smaller with 18 to 23 participants included in analyses (31, 39, 43, 44).

Finally, study participant heterogeneity is likely to influence results on many levels. It is conceivable that treatment effects of different psychosocial interventions will vary based on individual characteristics such as symptom burden and preferences. Conditions reflecting symptom burden and level of functioning are reflected in the included trials: participants in the studies with significant changes between the groups for positive symptoms measured by PANSS, were recruited from a psychiatric rehabilitation ward (29), from a day care center (29) and from a residential center (31), whereas participants in the studies with non-significant changes were hospitalized (39, 43, 44). Nevertheless, analyses of 27 hospitalized participants showed significant changes between the groups for positive symptoms, and not for negative symptoms, measured by a questionnaire (45). Recommendations for further research include dividing participants into subpopulations as well as investigating whether the effectiveness of AAI varies based on severity of symptoms, demanding a relatively high number of participants.

Diverging results have also been documented in previous SRs on related topics. As an example, a SR on dog presence and therapeutic alliance stated that half of the studies showed effect. Heterogeneity in study characteristics was described as an important limitation (62). Some of the results in our SR, however, contrasted earlier findings. In the SR by Hawkins et al. (25) including RCTs on AAI in general for individuals diagnosed with schizophrenia and related disorders, no improvement regarding quality of life was reported. One of the studies included in our SR indicated the opposite (32). This underpins that the field is under continuous development.

One of the purposes of this SR was to examine somatic effects, which were directly assessed in some of the studies through examination of physical skills and measurements of biochemical markers (30, 39). However, heart rate, blood pressure, HbA1c and lipid levels have not yet been investigated for adults diagnosed with schizophrenia and related disorders participating in DAI. This will be of importance as the population is at high risk of metabolic syndrome (63). Beneficial effects on cardiovascular risk factors have been associated with dog companionship or therapy for varied populations, but it is stated that further research is needed (64).

Another outcome measure especially relevant for further investigation is motivation. No significant changes were found for motivation related to treatment as a measure of quality of life, but this was only investigated in one study (43). Adherence can also function as an indicator of motivation. The adherence was significantly higher in the intervention group compared with the control group in one study (39). However, the results remain inconclusive. In addition to stating of reasons for non-adherence and lost to follow-up, validated instruments such as IMI-SR (65) may provide valuable information in further studies.

Significant worsening of non-personal social behavior within the intervention group was reported in one study (44). According to the authors, the intervention was not directed at these aspects and a similar trend was seen within the control group. Overall, prevention of negative consequences should have high priority. Examination by specialists in veterinary behavioral medicine was one of the preventive interventions described in one of the studies (39). Another example was inclusion of another dog in the later stages to reduce anxiety and grief due to removal of the dog at the end of the study (41). In a SR specifically addressing the benefits and risks associated with AAI, allergies, infections and accidents were described as the major risk factors. It was stated that these factors were outweighed by benefits (14). Animal welfare was only mentioned specifically by one of the included studies in our SR (39). Methods for overall safety, prevention of negative consequences and welfare for both participants and animals are of importance to describe in further articles. Development of interventions must be performed in accordance with guidelines for safety and welfare, for example from IAHAIO (13). A specific recommendation for further research is to describe evaluations regarding signs of stress in the participating dogs. In addition, a predetermined plan of action in case of negative consequences is of importance to include.

In addition to abovementioned limitations regarding consistency, a low number of participants led to imprecision. Furthermore, three outcome measures were associated with indirectness due to surrogate measures or use of inappropriate measurement methods. Risk of bias was categorized as high or serious for most of the included studies. Some factors that entailed the risk were missing outcome data, deficient blinding of personnel who assessed the measurements and risk of confounding. However, some of the factors causing risk of bias could not be avoided due to the nature of the interventions.

Exclusion of articles written in other languages than English and Scandinavian led to risk of selection bias in the review process. Lack of access to potentially relevant studies may also have caused bias. Due to lack of variables, such as confidence intervals, findings were reported as significant or non-significant. Such reporting is against the principles of Cochrane, and it must be emphasized that lack of evidence is not the same as lack of effect (66). Therefore, both significant and non-significant findings must be interpreted with caution.

Although inclusion of other designs than RCTs led to lower quality of evidence, these studies proved valuable in this SR through presentation of outcome measures not synthesized in a previous SR (25). Furthermore, inclusion of recent published studies led to novel insight on the topic. Isolated assessments of anhedonia, social function (including social adaptive function), communication and interaction skills (including non-verbal communication), lower body strength, mobility, agility and cognitive function, were among the outcome measures that expanded the knowledge. Specific examination of dog-assisted interventions increased the directness and complemented more general reviews on related topics.

Summarized, the findings suggest that DAI may have an effect on a range of symptoms and features associated with severe psychotic disorders. However, the findings must be interpreted with caution. Due to several knowledge gaps, it is challenging to state specific implications for policy and practice. The trade-off regarding potential benefits and potential harms are important. Based on available data, we consider the potential benefits of DAI to outweigh risk of harmful effects, given that all required precautions are taken. A potential negative consequence was described by only one of the included studies, and the causality of the finding remained uncertain (44). A specific implication for practice, which must be emphasized, is the necessity of development and implementation of interventions in accordance with guidelines for safety and welfare. The lack of data regarding animal welfare assessments is considerable, and the area is overall described as under-researched (67). In addition to the specific recommendations for further research presented in the paragraphs above and in Supplementary Table 39, reduction of bias and increase of quality will be essential. Accordingly, there is a specific need for carefully designed RCTs. This is particularly justified by the risk of confounding associated with NRS.

The included studies indicate potential effects of dog-assisted interventions for adults diagnosed with schizophrenia and related disorders, mostly beneficial. However, the results must be interpreted with caution due to methodological limitations such as low number of participants, heterogeneity among study design and included participants, and risk of bias. Findings of both significant and non-significant results are in accordance with reviews on animal-assisted interventions in general. Importantly, inclusion of several study designs and novel trials enabled synthesizing of outcome measures not covered by previous reviews. Some of the results, such as significant improvement for quality of life, contrast earlier findings. This underpins that the field is under continuous development, and further examination of causality is warranted. Recommendations for future research include factors such as calculation of effect sizes, development of more standardized programs, and investigation of effects related to motivation and somatic effects.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

MT designed this systematic review with contributions from all co-authors, wrote the original draft, and all authors participated in revision. EJ and MT extracted data and graded the quality. MT and SS assessed risk of bias. All authors participated in screening and selection of articles and approved the final manuscript.