2.2.4. Publication Bias

When publication bias with respect to agitation, four studies were within the 95% confidence interval and were plotted to the left of the overall effect estimate (Figure 4A). When publication bias with respect to depression and QOL with respect to the effect of AAI and PRI was assessed (Figure 4B,C), all plotted dots were within the 95% confidence interval.

**Figure 4.** Funnel plots used to assess the existence of publication bias in the included studies. Publication bias of (**A**) agitation, (**B**) depression, and (**C**) quality of life.

#### **3. Discussion**

Currently, more than 90% of dementia patients suffer from BPSD [39], which poses major difficulties to both dementia patients and their caregivers. The type of BPSD varies according to dementia type, stage of the illness and various other factors. Particularly, patients of frontotemporal lobar degeneration (FTLD) show more prominent behavioral variants such as disinhibition, impulsivity, aggression, and personality change than those with other types of dementia [40–42]. Another study demonstrated that patients with dementia with Lewy bodies (DLB) present hallucinations and aberrant motor behavior (AMB) more so than Alzheimer's disease (AD) patients [43,44]. An increased rate of anxiety, depression, and psychosis may occur in vascular dementia (VD) [40,43,45]. Depression and agitation are the most common symptoms affecting various dementia patients. Furthermore, it is known that agitation, apathy, disinhibition, irritability, and motor dysfunction become serious as dementia progresses. In particular, depression and anxiety become more severe in the moderate stage of dementia [46–48]. In the early stages of dementia, apathy mainly appears, which is one of the first symptoms of the various forms of dementia. Apathy is a dangerous barrier that affects social interaction and activities of daily living due to lack of interest, enthusiasm, and apathetic response to interpersonal communication [49]. These psychological and behavioral changes from the early stages of dementia can affect aspects of BPSD such as depression and anxiety more seriously as dementia progress. Although BPSD, which varies depending on the type and progression of dementia, contains a range of important symptoms that affect the quality of life, stress, and prognosis of dementia patients and their caregivers, there is little of interest in and study on nonpharmacological interventions to treat BPSD. Thus, we performed a meta-analysis to investigate the effect of AAI and PRI—one of the

nonpharmacological interventions using animals— on agitation, depression, and QOL in dementia patients [15,26,27].

The meta-analysis of the effects of AAI and PRI on agitation showed a medium effect size of 0.70 (Figure 1). Three studies that utilized AAI and two studies that utilized PRI were included in the meta-analysis. The studies that used AAI reported larger effect sizes than those that used PRI, but AAI and PRI were not found to significantly affect agitation overall [23,24,35]. Our result contrasts with the results of a previous study which showed an alleviation in the agitation. However, since the level of evidence for the randomized controlled trials (RCTs) in previous studies was very low, we thought that the opposite results were obtained. Accordingly, our results support the suggestion of previous studies that the level of evidence is low [32].

The meta-analysis of the effects of AAI and PRI on depression showed a medium effect size of −0.47 (Figure 2). Three studies that used AAI were included, and two reported that this intervention strategy reduced depression [23,24,35]. Two studies that used PRI were included, and these showed a medium effect size [36,37]. AAI and PRI were found to significantly reduce depression, which serves as evidence that AAI and PRI are effective at reducing depression in dementia patients (*p* < 0.001).

The meta-analysis of the effects of AAI and PRI on QOL showed a small effect size of 0.13, but the results were not statistically significant (*p* > 0.05) (Figure 3). Two studies used AAI, and both reported that these interventions improved QOL [11,24]. One study used PRI, and reported that this intervention did not significantly affect QOL [36]. The meta-analysis results showed that AAI and PRI did not significantly affect QOL, which supports previous findings [32].

The present study analyzed the effects of AAI and PRI on BPSD and found that these interventions did not affect agitation or QOL but significantly reduced depression. It is well known that the brain with depression in dementia has reduced connectivity on amygdala and emotion control regions [50,51]. AAI and PRI provide an emotional effect and a and sense of closeness to dementia patients [52], which may the reduced amygdala connectivity in dementia patients. In addition, AAI and PRI could have a positive effect on hippocampus in the brain with depression through activities that require memory, such as checking the health of animals, walking, and feeding. On the other hand, the agitation-related connectivity is the orbital frontal cortex and anterior cingulate cortex, which is a region that has little association with emotional support obtained through activities with animals. Thus, AAI and PRI did not show a significant effect in agitation. Although AAI and PRI have been effective in improving depression, it is difficult to dramatically relieve all BPSD symptoms. Moreover, it is known that BPSD is specifically related with the patient's low of QOL [53]. Therefore, in this study, it is considered that AAI and PRI were difficult to significantly influence QOL. A previous meta-analysis reported that AAI do not affect activities of daily living, depression, agitation, QOL, or cognitive function. In addition, a number of limitations are associated with interventions involving the use of living animals: patients may be fearful of or allergic to animals, animals may provoke falls in vulnerable patients, and animals may pose an infection risk to patients [32]. Moreover, there are a number of difficulties associated with managing animals—they need to be fed, produce feces, and may smell. However, it is clear that AAI can enhance the emotional wellbeing and QOL of dementia patients. Although robotic animals cannot evoke the same variety of emotions and sensations as living animals, they are easier to manage and could aid patients wherever needed. Subsequent studies should additionally examine the impact of living animals and robotic animals on the emotional wellbeing, cognitive function, and physical ability of dementia patients. Furthermore, patients should be followed-up to investigate the efficacy of these interventions in slowing the progression of dementia.

Several studies have suggested that psychiatric symptoms such as depression and anxiety are associated with dementia and cognitive impairment [54–56]. Indeed, patients with dementia have an increased risk of major depression, and many suffer from anxiety [57,58]. Interestingly, amyloid-beta (Aβ) burden and tau-related pathology are known to worsen in Alzheimer-type dementia with depression [55,59]. In addition, depression and agitation are causative factors of sleep disorders, and they can promote the development of dementia by inhibiting Aβ clearance and inducing systemic

inflammation [60–63]. Therefore, it is important to alleviate the psychological symptoms of dementia patients. In this study, we confirmed that AAI and PRI can relieve the psychological symptoms of dementia patients. Several mechanisms by which AAI and PRI may affect BPSD have been proposed. First, AAI and PRI affect hormone levels. Previous studies consistently reported that dog-raising people exhibit higher levels of oxytocin, a hypothalamic neuropeptide [64,65]. Oxytocin is closely related to cognitive function, depression, agitation, and social communication and has been proposed as a pharmacological intervention for neurobehavioral disorders in patients with prefrontal dementia [66,67]. In addition, it has been reported that animal owners exhibit reduced cortisol levels [68]. In AD, cortisol levels substantially increase and this steroid hormone elicits neurotoxic effects in the hippocampus and thus exacerbates Aβ pathology and contributes to cognitive impairment [69]. Therefore, AAI may improve BPSD by increasing oxytocin levels and reducing cortisol levels. Furthermore, the relationship between loneliness and depression is well established, and loneliness has been reported to promote Aβ deposition in the brain of AD patients [70,71]. In addition, loneliness is known to contribute to cognitive decline by lowering cognitive reserve [72]. Surprisingly, AAI is known to reduce the loneliness of residents in long-term care facilities [73]. Therefore, AAI and PRI may effectively reduce loneliness and depression in dementia patients.

Second, it is possible that AAI and PRI modulate brain structure and functional connectivity. Patients with dementia exhibit atrophy of the hippocampus and entorhinal cortex, areas of the brain associated with emotional and spatial memory [74]. In addition, late-stage dementia is associated with dysfunction of the amygdala and cerebral cortex [75,76]. Accordingly, patients with dementia have problems with language, reasoning, emotions, and social behavior. Furthermore, atrophy of the hippocampus and cerebral cortex affects the functional connectivity of frontotemporal and limbic circuits involved in depression and mood regulation [77]. Strikingly, emotion-related brain areas may be affected by dementia patients' relationship and emotional stability. Indeed, improvements in executive function, social skills, mood regulation, learning, memory, and attention were noted in patients receiving cognitive rehabilitation therapy through various AAI [52]. In addition, in children with ADHD, AAI had a calming effect, increased motivation, improved cognitive function, and promoted socialization [78]. It is thought that interaction with a therapy animal enhances functional connectivity between the frontotemporal and limbic systems. Moreover, having to look after an animal and remember to perform tasks such as feeding it is thought to improve memory and learning ability and attenuate hippocampal and cortical atrophy. Social interaction is possible through relationships and walking with animals, and through group meetings, depression will be alleviated. Although the neurological mechanisms underlying the effects of AAI and PRI have not been fully elucidated, accumulating evidence suggests that AAI and PRI can effectively improve BPSD.

Although a number of previous studies have also investigated living- and robotic- animal-assisted interventions for patients with dementia, our study has a number of strengths [31–33]. First, we comprehensively investigated the effects of interventions involving living and robotic animals and, for the first time, compared the effects of AAI and PRI on BPSD. Second, we demonstrated trends in research in this field and confirmed that more research is now being conducted into interventions involving robotic animals for dementia patients. Third, two reviewers independently identified articles that met the inclusion criteria, and a high level of inter-rater agreement was noted. Fourth, we focused on BPSD and dementia. Although AAI and PRI are known to affect various symptoms of dementia patients, we conducted a literature search and meta-analysis focusing on BPSD. Finally, it is difficult to distinguish between mild cognitive impairment (MCI) and dementia patients unless a neurological examination is performed to definitively diagnose dementia. In this study, we aimed to confirm the effect of AAI and PRI in individuals who had been diagnosed with dementia, not MCI.

Nevertheless, our study has a number of limitations. One limitation of the meta-analysis is the small number of included studies, which shows that there is a lack of literature relating to AAI and PRI for dementia patients. In addition, we only selected studies published in peer-reviewed journals and did not include any grey literature, which may have introduced publication bias. Third, we were unable to identify specific subgroups of dementia patients who may benefit most from AAI and PRI. Finally, we searched only a few English language databases, so some relevant studies may have been missed.
