The research question for this scoping review is “What is known about the impacts of low-cost, familiarly and realistically designed interactive robotic pets (i.e., the JfA robotic dog and cat) for older adults and people with dementia?”

Published articles and grey literature were identified and searched in the following electronic databases: CINAHL, Web of Science, Scopus, MEDLINE via Ovid, and PsycINFO via Ovid. All relevant literature that were written in English, regardless of methodological quality, were included. Since the JfA robotic pets were only developed in 2016, only studies published after 2016 were included. The search strategies were developed in consultation with a research librarian based on the Population, Concept, and Context (PCC) framework that is recommended by the Joanna Briggs Institute for scoping reviews (Textbox 1). The full search strategy can be found in Multimedia Appendix 1. To cover the breadth of available literature and to ensure that the search was comprehensive, searches were also conducted on Google Scholar and through forward and backward citation tracing. The search was initially conducted in May 2020. To maximize the currency of this review [36], an update of the search was conducted in September 2020.

The selection of studies followed a two-stage screening process. Two independent reviewers (authors WQK and FXHA) were involved in the screening process. Any nonconsensus or discrepancies were discussed and resolved among both reviewers and with author DC, as necessary. First, the titles and abstracts of identified articles were independently screened. We anticipated that information regarding the specific type of robotic pet (ie, the JfA robotic cat and dog) may not be mentioned in the title or abstract of publications and may only be available in the body of the text. Therefore, all studies were included if they met the following inclusion criteria based on the PCC framework: had any type of primary study; used a robotic cat or dog as an intervention; included older adults 60 years or older, or people with dementia; and were published in the English language. The exclusion criteria included if they were noninterventional studies such as expert opinion and commentaries, used any other robotic pets such as Pleo or AIBO, did not include older adults (ie, younger than 60 years), and were published in languages other than English. If these criteria were unclear in the title and abstract screening, they were included for full-text screening. Second, the full texts of included articles were reviewed. Studies that employed the JfA robotic pets were included, and studies using any other robotic pets such as the Justocat and NeCoRo cat were excluded. Any disparities were discussed and resolved. A bibliographic reference management tool, EndNote, was used to ensure that all articles were systematically accounted for. The search strategy was recorded using a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flowchart (Figure 2).

A standardized data extraction form was created using Excel (Microsoft Corporation). The data that were extracted included authors, country of the study, research design, research setting, participants’ demographics, sample size, intervention delivery, positive impacts, and negative impacts. Authors of included studies were contacted as necessary to attain additional information. Both reviewers (WQK and FXHA) charted the data independently before making comparisons afterward. Both reviewers discussed to collate the extracted data into categories and refined them to develop the final themes. The PRISMA-ScR (Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews) checklist (Multimedia Appendix 2) was used to guide the reporting of the results [37].

Although quality appraisal is not necessitated for scoping reviews, it has been recommended to evaluate the methodological integrity of included articles [38]. Two reviewers (WQK and FXHA) independently appraised the quality of the included studies before meeting to discuss any discrepancies, which were resolved through discussion and a consensus was reached.

Qualitative studies and the qualitative strand of the mixed method study were appraised using the Critical Appraisal Skills Program qualitative checklist [39]. The research aims and rationale of all studies (n=7) were clearly stated. With the exception of 1 study [40], most studies confirmed that ethical approval was obtained from a relevant research ethics committee. Most had appropriate research designs (n=4) [40-43] and recruitment strategies (n=5) [40,41,43-45]. However, the data collection and analysis methods were not clearly described in 4 studies [42,46,47]. These factors subject the studies to assessor bias and reporting biases [48]. Emails were sent to the authors to request for more information; however, no responses were received. Most studies (n=6) did not provide sufficient information to illustrate if the relationship between the researchers and participants were adequately considered [40-43,45,47].

The National Institutes of Health (NIH) quality assessment tool for pre-post studies [49] was used to appraise the quantitative study and quantitative strand of the mixed method study. The tool contains 12 questions to guide reviewers’ judgement of whether a study is of “good,” “fair,” or “poor” quality. The quality of these studies were rated as poor and fair, respectively. In the mixed method study by Marsilio et al [45], it was unclear whether all eligible participants were enrolled, which subjected it to selection bias. In addition, the intervention was not clearly described, suggesting the potential for information bias. The other study by Tkatch et al [50] had a significant attrition rate. Furthermore, both studies did not state whether assessors were blinded, which raised concerns about reporting biases [45,50].

Finally, the Authority, Accuracy, Coverage, Objectivity, Date, Significance (AACODS) checklist [51] for appraising grey literature was used to evaluate the quality of McBride et al’s [46] article. This article did not have a clearly stated aim or research design. An email was sent to the authors requesting more information, and an author clarified that the study was unstructured, and there was no additional information beyond what was presented in the article. Hence, this article was rated to be of poor quality. The full quality appraisal tables can be found in Multimedia Appendix 3.

Overall, the quality of reporting in the included studies varied from poor to good, with most classified to be of poor to fair quality. Nevertheless, all studies were included in this scoping review, as the intention of this review is to identify the breadth of literature in this topic (Table 1).

Table 1

Characteristics of included studies.

The sample sizes in 8 studies ranged from 1 to 216 and included a total of 296 participants. It was not possible to ascertain the sample size in 1 study [41]. Most studies (n=6) were conducted with older adults with dementia [40-43,45,47]. However, 1 study also included older people with learning disabilities [41]. Healthy older adults were the participants in 2 studies [44,50]. In the remaining study, participants were older residents in a nursing home. However, there was no information on their ages or diagnoses [46]. Studies were conducted in participants’ homes (n=5) and in long-term care settings (n=4).

The majority (n=5) used the JfA robotic cat [40,42,43,45,47], while the others (n=4) employed both the robotic cat and dog [41,44,46,50]. Only 1 study offered participants’ their choice of robotic pet (ie, cat or dog) and reported no differences between the type of pet to the intervention outcomes [44]. The intervention duration ranged from 2 weeks to 6 months. The majority (n=9) delivered the robotic pet as a one-to-one intervention. Only 1 delivered the intervention both individually and communally [41]. Most (n=5) provided the robopet to participants on a full-time basis [42-44,50]. In 1 study, their use progressed from structured 1-2 hour sessions during the first 2-3 months to full-time use by the third month [41]. Finally, 2 studies reported intervention delivery on a weekly basis, between 1-3 times each week [41,46].

In most studies (n=7), minimal facilitation or instructions were provided by the researchers to guide intervention delivery with the robotic pets to allow their use to be scaffolded naturally [40-44,47,50]. Among studies that provided information about intervention delivery during the research, 3 reported facilitation by formal caregivers [41,45,46]. In 1 study, staff placed the robotic pet in the resident’s arm, talked about it, and then left the resident alone with it [45]. It was also made available during other times when residents asked for it or when the nurses were motivated to use the robotic pet with residents. Another study reported that, although the robotic pets were available in communal areas for unfacilitated interactions, structured group sessions with the robotic pets were also delivered by staff [41]. Finally, difficulties integrating the use of the robotic pets into nursing routines were reported in 1 study [46]. As such, nurses relied on therapeutic recreation staff to use them with nursing home residents [46].

The positive impacts included improved mood and affect, improved communication and interaction, companionship, and other well-being outcomes (Table 2).

Issues and concerns related to the use of the robotic pets included misperception and attachment, no impact or negative impacts, and practical issues.

This is the first scoping review to identify and synthesize the evidence on the delivery and impact of low-cost, familiarly and realistically designed robotic pets for older adults and people with dementia. The majority of the included studies in this review were conducted in long-term care facilities and in participants’ homes, and most employed the JfA robotic cat.

Overall, the positive impacts of the JfA robotic pets related to several psychosocial domains. Positive impacts included improved mood and affect, communication, social interaction, and companionship; these benefits resonate with findings in reviews that investigated the effectiveness of other social robots and robotic pets for older adults and people with dementia [13,14,16]. However, the impacts on other domains, including loneliness, resilience, and purpose in life, were less investigated; in this review, only 1 study that focused on cognitively healthy older adults reported on such outcomes [50]. This corresponded with findings from a review paper that investigated the use of social robots for older people [52] and found that only 3 studies reported outcomes relating to loneliness among healthy older adults. Similar to studies using other robotic animals, the interactivity of the JfA robotic cat and dogs have been described to facilitate users’ communication and interaction with the pet and with other people. Paradoxically, the interactive features of the JfA robopets caused distress among a few participants with dementia. Such issues have been reported previously, where users were disturbed by sounds produced by another robotic pet [18,53-55]. Moving forward, there is a need for robot developers to consider the customizability of the robopets’ interactive features in accordance with users’ preferences.

The issue of affordability has been reported to impede the use of robotic pets in the real world [18,21,22,24]. The low-cost of the JfA robotic pets appeared to have an influence on intervention delivery and the conduct of research; with the exception of 1 study, all participants in this review received their own robotic pet for individual use. This is in contrast to findings from a systematic review, which found that higher-cost robotic pets have been shared among users and used more frequently in group settings [13]. The affordability of the JfA robotic pets was also cited by researchers as one of the influencing factors in the choice of robotic pet for their studies [40,41,43,50]. Cost appeared to have played a role in influencing the research method in one study, where individual robopets were provided to 216 participants to enable a statistical significant analysis of their impacts [50]. This strategy may be more challenging to implement with more expensive robots [16]. In addition, it is worth noting that there is a relatively sizeable body of anecdotal evidence, largely stemming from individuals’ reports of their experiences with this technology [56-59]. This might also be attributed to their affordability, which might have enabled more users to gain access to this technology as compared to other social robots that are more expensive. For example, although Paro is one of the most researched social robots, it has substantially less user-generated reports of its impacts. This could be because Paro is primary used in institutions [17], likely due to its cost, which renders it to be less accessible for individual users’ purchase. Individual ownership of the robotic pets may be viewed as a promising way to mitigate the pertinent issue of infection control, especially in light of the ongoing COVID-19 pandemic. A recent publication by Bradwell et al [60] reported that the acceptable levels of microbes on robopets, including one with antibacterial fur covering [17], exceeded an acceptable threshold after 20 minutes of use. Frequent and shared use of these robopets between different users can further increase the potential of infection transmission [27,28]. Hence, since the lower cost of the JfA robopets increases the affordability of individual purchases for each user, the corresponding risk of direct or indirect contact transmission of infections related to shared use may be ameliorated.

Issues related to use of the JfA robopets were identified. Like other interventions involving social robots, there were issues associated with use of this intervention. Some participants with dementia did not benefit from their use or demonstrated negative responses toward the robopets. For this population, the ethical challenge of deception also emerged [10], as some participants misperceived them as real animals or showed attachment toward them. These issues are not unique to the JfA robotic cat and dog, as they have been reported in other studies using other robotic pets [23,33,61]. The significance of these issues should not be discounted, as those who were more attached or misperceived the robopets belonged to a vulnerable population. However, from the standpoint of the capability approach, all humans, including people with disabilities, should be given the opportunity to achieve a threshold level of core capabilities to uphold the principle of social justice [62]. Therefore, in consideration that the pet robot may facilitate a user’s capacities that would be otherwise undermined, such as facilitating social interaction, this can be viewed as enabling technology with greater benefits than risks [63]. In addition, formal and informal caregivers should also explicitly consider upholding this principle, particularly when delivering the robotic cat. When introducing this technology to users, they should introduce it as a robotic pet and refrain from referring to it as a real animal [63]. The understanding of potential issues such as jealousy and attachment may also guide future implementation and inform future robot development to ensure robustness of the technology.

Users’ responses toward the JfA robopets appear to be related to their profile (ie, preference for or experience with animals). Participants who did not respond or had negative responses to the JfA robopets were reported to not like animals. This aligned with findings from other studies that highlighted that multilevel stakeholders including people with dementia [17], family members [18], and staff [22] who liked animals had positive perceptions and reactions to robotic pets. Therefore, before considering the use of the JfA robopets to address the psychosocial needs of older adults or people with dementia, care providers should consider users’ preferences for animals, as well as their preferred type of robotic animals, to maximize the appropriateness and meaningfulness of the intervention.

There are a number of strengths underpinning this work. First, the methodological framework used throughout the scoping review process was transparent and rigorous. The screening and data extraction process involved two independent reviewers, which reduced the risk of reviewer bias or article selection bias. Both reviewers met at regular intervals and discussed and resolved all discrepancies. Second, this paper discusses the pragmatic aspects relating to intervention delivery and the conduct of research using the JfA robotic pets, which can serve as useful considerations for researchers or users who are keen to further explore the use of this technology. However, there are limitations of this review. Articles that were published in other languages were not searched or included in this review. As non-English studies were excluded from this review, relevant studies may have been missed.

This scoping review has mapped out current evidence on the use of and impact of realistic and familiarly designed low-cost robotic pets for older adults and people with dementia. Our review contributes to the evidence base that is necessary for more widespread awareness about the potential utility of these low-cost robotic pets to address the psychosocial needs of older adults and people with dementia, as both the positive impacts and issues related to their use largely resonate with research conducted with several other robotic animals. The affordability of these robopets appear to have an influence on intervention delivery. They also appear to have the ability to uphold the distributive justice of innovation dissemination; these are especially relevant in light of the COVID-19 pandemic, where there is an increased emphasis on infection control and equal access. However, more rigorous effectiveness trials are required to confirm their positive impacts. Future studies should also consider comparing the intervention effects of the JfA robotic pets with other robotic pets. It is also important to ascertain the design preferences of older adults and people with dementia to facilitate the development of future user-centered interventions using robotic pets.