More than 6 million individuals in the United States are living with Alzheimer disease and related dementias (ADRD) and it is attributed as the cause of death for 1 in 3 individuals over the age of 65 [1]. In the past 2 years, deaths attributed to ADRD have increased by 16% and research has projected that the number of people living in the United States with ADRD will triple by 2060 [1].

Individuals living with ADRD require prolonged and complex care that is primarily managed by informal caregivers. Informal caregivers are unpaid, nonprofessionals that provide care and typically include family and friends. There are an estimated 11 million caregivers providing care for people living with ADRD in the United States and they provide approximately 15.3 billion hours of unpaid care valued at nearly US $257 billion [2]. Caregivers report being undertrained, under-supported, and under-resourced to perform their caregiving role. Although caregivers can experience positive outcomes related to caregiving, the imbalance of caregiving demands and supports is often associated with mental, physical, and economic challenges that can lead to significant consequences for caregivers and the individual living with ADRD, such as caregiver stress, burden, depression, and morbidity [3-5].

To address these suboptimal caregiver outcomes, the US National Institute on Aging and other national advisory panels have highlighted the development and testing of technology-based interventions for caregivers of people living with ADRD as a key priority [6-8]. For example, the 2015 Alzheimer’s Research Summit highlighted the need to “test the use of technology to overcome the workforce limitations in the care of older adults with dementia as well as providing caregiver support and education.” [7]. In response, researchers have developed numerous information technology interventions such as mobile apps and websites to support ADRD caregivers across a range of domains including caregiver education, self-care support, support for managing behavioral symptoms of dementia, and virtual peer support groups [9-11]. Several systematic reviews and recent meta-analyses report that these technology interventions can improve outcomes for caregivers, such as increased self-efficacy and reduced ADRD caregiver burden, stress, and depression [3,12-19]. These reviews also suggest that effective interventions offer multiple components, tailored options, and social support [9-11].

However, a significant gap in existing interventions is that most focus only on the primary caregiver, even though most people living with ADRD receive care from more than 1 caregiver—a care network—with varying degrees of involvement [20-24]. Currently, caregivers face significant unmet needs regarding support for communicating and coordinating across the care network including sharing information, maintaining situation awareness, distributing responsibilities, scheduling, and managing caregiver hand offs [20-24].

Although some mobile apps exist that allow caregivers to share information, they are limited in their functionality, quality, and potential to meet the specific needs of ADRD caregivers [25]. A recent review of mobile apps for caregivers of people living with ADRD available on the US market identified 2 mobile apps that support shared communication and coordination [25]. According to study findings, one of those apps did not function consistently and received a quality rating of inadequate as indicated by the Mobile Application Rating Scale (MARS). The second app received an overall quality rating of minimally acceptable quality according to the MARS but scored lower than average on subjective quality. A similar study conducted in 2018 by Wozney et al [26] identified 3 mobile apps that connect a primary caregiver to other members of the care network. One of the apps identified is no longer available on the US market. The other 2 apps identified are not specifically designed to meet the needs of caregivers for people living with ADRD and are limited in their function (eg, only provide a shared calendar).

To address the current gaps in existing interventions, we developed CareVirtue, a progressive web application developed in React to support and connect ADRD care networks that can be accessed via a web browser on any device with a data connection. CareVirtue seeks to address the current gaps in existing ADRD caregiver support technologies through a high-quality, user-centered, ADRD caregiver–specific multicomponent technology to support communication, coordination, and connection among care networks. CareVirtue was initially inspired by an online support group for people newly diagnosed with ADRD and their caregivers in which support group members expressed an unmet need for tools to support communication and coordination among the care network. The need for CareVirtue was further supported by several findings from foundational research on care networks. First, findings from Block et al [27] suggested that primary caregivers require technologies to communicate and coordinate among the care network, that they try to adapt existing technologies (eg, email, messaging) to meet their needs, and that adaptation requires additional time and effort. Further, Ponnala et al [20] found that for primary caregivers, the currently under-supported communication and coordination among the care network increases their caregiving demands. Moreover, Tang et al [24] highlighted the consequences caregivers experience with under-supported care network communication and coordination, including maintaining situation awareness among caregivers, missing care information leading to potential patient harm (eg, missing a medication dose), and miscommunication leading to care network tensions or conflict.

Collectively, this prior research provides the foundation for CareVirtue. CareVirtue’s design honors the person-centered care model for people living with ADRD and their caregivers by (1) treating people living with ADRD as individuals with unique needs; (2) seeing the world from their perspective; and (3) creating a positive social environment in which the person living with ADRD and caregiver can experience relative well-being and quality of life. CareVirtue was designed and developed through consistent, iterative user input across multiple stages of usability testing coupled with expert evaluation. CareVirtue was specifically designed to encapsulate the foundational principles of person-centered care through the following features. Multimedia Appendix 1 provides a detailed walkthrough of CareVirtue features, which are given in brief in Textbox 1.

In this study, we used a mixed methods approach to evaluate CareVirtue’s feasibility across the following research questions:

We conducted a feasibility study over a period of 8 weeks with the purpose of demonstrating CareVirtue’s acceptability and feasibility among care networks of people living with ADRD. This study reports one aim of a larger project, which has 2 specific aims. The first aim is the focus of this study. The second aim is to leverage the CareVirtue data generated by this feasibility study to develop an intelligent caregiver assistant (R41AG069607). The larger sample size and longer study duration than is typical for feasibility studies is due to adjustments made to achieve the second goal [28].

Participants were recruited between February and June 2021 through multiple community sites in Wisconsin and Southern California as well as through the Wisconsin Alzheimer’s Disease Research Center. Advertisements for study participation were distributed via email, social media, and newsletter posts. Interested individuals contacted the study team via email or phone and were subsequently phone screened for the following eligibility criteria: self-identified primary caregiver of a person living with ADRD, at least 18 years of age, English speaking, daily internet access, and shares caregiving information/responsibility with other caregivers.

Eligible participants were scheduled for a 1-hour enrollment visit via videoconferencing software. During the enrollment visit, a study team member obtained informed consent from the primary caregiver and from the associated person living with ADRD (ie, the care recipient). If the person living with ADRD did not have decisional capacity to consent, the primary caregiver could consent on his/her behalf if he/she was the legally authorized representative. Next, the study team member administered a pretrial demographic survey, helped create the CareVirtue account, and provided a walk-through of CareVirtue’s functionality. Primary caregivers selected and invited secondary caregivers (to form a care network) at their own discretion. Once secondary caregivers were invited, they were separately contacted via email to electronically obtain informed consent.

Following enrollment, participants used CareVirtue for 8 weeks. During the use period, we administered a weekly survey starting 1 week after enrollment to assess caregiver workload and confidence using CareVirtue. If the survey was not completed, a follow-up reminder was sent the following day. At the completion of 8 weeks of use, we conducted a posttrial visit with the primary caregiver participant via Zoom where we administered posttrial surveys and conducted a semistructured interview as described in the following section. Primary caregivers were provided with a US $150 e-gift card at the end of the study period. Secondary caregivers contacted the study team if they were interested in participating in the postuse survey and interview. Secondary caregivers received US $15 for completion of the postuse survey and US $25 for completion of the 30-minute postuse interview.

Research ethics approval was granted by the Institutional Review Board at the University of Wisconsin-Madison (Protocol #2020-1035).

We enrolled 51 primary caregivers of people living with ADRD (Table 1) and 61 secondary caregivers to use CareVirtue during the study period. Care networks ranged from 1 to 8 members. Primary caregivers were predominantly female (38/51, 75%), White (44/51, 86%), married (37/51, 73%), college educated (36/51, 71%), and were, on average, 60.3 (SD 9.8) years of age. Care recipients were also primary female (34/51, 67%) and White (45/51, 88%), with an average age of 79.2 (SD 10.6). Care networks were located in both Wisconsin (29/51, 57%) and California (19/51, 37%), with 18% (9/51) living in a rural area. During the study period, 4 primary caregivers dropped out because of care recipient death (n=2) and personal situations (n=2). We were unable to reach 6 primary caregiver participants for the posttrial visit. A total of 12 secondary caregiver participants completed the postuse survey.

Figure 5 displays boxplots (across care networks) for each of the 8 usage characteristics. The average (SD) was 18.3 (22.4) log-ins, 32.5 (46.5) journal posts, 5.3 (13.2) journal post replies, 10.6 (28.5) calendar events, 2.2 (2.1) secondary caregiver invites sent, 2.2 (2.1) secondary caregiver invites accepted, 6.1 (0.4) care guide sections created, and 0.6 (1.9) resources accessed. The log-in data are not fully representative of actual use because users could remain logged in to CareVirtue depending on their preference to log out.

Table 2 presents the centroid (the mean values across all care networks in that cluster), cluster size, and a cluster label for each of the 8 clusters. The 8 clusters could be further reduced into 3 primary groups according to the degree of engagement with the platform. There was heterogeneity in how the most engaged care networks interacted with the platform; for example, 2 care networks made heavy use of the calendar feature (average of 141 events) with few posts (average of 41), while another care network made heavy use of the journal (257 posts) with only 1 calendar event.

We used the NASA TLX score to assess usability and usefulness in terms of caregivers’ perceptions of their caregiver workload. Figure 6A displays boxplots of the total NASA TLX score for each week from week 0 (before the study began) to 7 (the final week of the study). The average NASA TLX score increased over the duration of the study (P=.02; r=0.79, 95% CI 0.65-0.87). However, at an individual level the NASA TLX score only increased over the duration of the study for 3 primary caregivers, decreased for 2 caregivers, and did not change for the remaining 29 primary caregivers (17 were excluded due to missingness).

We used a confidence scale to assess CareVirtue usability. Figure 6B displays boxplots of the total confidence score for each week of the study. The average confidence score increased significantly over the duration of the study from a low of 7.8 in week 2 to a high of 8.9 in week 7 (P=.004; r=0.91, 95% CI 0.84-0.95). At an individual level, 7 primary caregivers saw a statistically significant increase (P<.05 in all cases; see Multimedia Appendix 5 for precise P values) in confidence, while the remaining 17 remained stable (27 were excluded due to missingness).

We also used the SUS to assess CareVirtue usability and the Behavioral Intention Scale and Perceived Usefulness Scale to assess usefulness. Multimedia Appendix 6 displays histograms for the Behavioral Intention Scale score, SUS score, and the perceived usefulness score for primary caregivers. The average (SD) was 3.3 (1.2) for the Behavioral Intention Scale and 72% (26/36) of primary caregivers reported an average score of at least 3, indicating an above average intention to use. The average (SD) was 81.8 (12.8) for the SUS score, indicating “good” usability and 3.4 (1.0) for perceived usefulness, suggesting above average perceived usefulness. For secondary caregivers the average (SD) was 2.9 (1.1) for the Behavioral Intention Scale (6/11, 55%, had an average score of at least 3), 77.9 (13.7) for the SUS, and 3.4 (0.8) for perceived usefulness (Multimedia Appendix 7).

Through analysis of the qualitative interviews, we identified 10 categories related to usefulness and 4 categories related to opportunities to improve usefulness (Table 3). Participants described CareVirtue as facilitating connection, exploration, and awareness. In addition, CareVirtue allowed for documentation and tracking of daily experiences and enabled emotional catharsis by facilitating the capture and review of significant moments in their relationship with the person living with ADRD. Additional facilitators of usefulness described by participants included centralization of information, coordination across the care network, and the privacy afforded by controlling user permissions. Participants further explained that using CareVirtue made them feel supported and reduced feelings of being overwhelmed. We also identified opportunities for improvement, which included increasing engagement such as creating a CareVirtue user support group; adding customization such as additional emoji options for journal posts; refining navigation such as expanding search types; and additional functionality such as document upload and storage.

The following sociodemographic characteristics were associated with the number of posts: retired (average of 59.5 posts for retired caregivers as compared with 16.9 for nonretired caregivers), income (13 posts for those reporting >US $100K and 55.4 for those reporting <US $100K), relationship to care recipient (18.7 posts for child and 56.4 for partners/spouses), and living situation (44.7 for those who live with the care recipient and 13.1 for those who do not). Older care recipients were associated with fewer posts (r=–0.33, 95% CI –0.55 to –0.06).

The following workload characteristics were associated with the number of posts: NASA TLX score representing the perceived workload associated with the caregiving role from weeks 2 to 7 (r=0.37-0.46) and the total hours caregiving (r=0.38, 95% CI 0.11-0.60). In other words, higher perceived mental workload associated with the caregiving role and a greater number of hours spent caregiving were associated with more journal posts.

Regarding usability, we found that a higher SUS score (r=0.40, 95% CI 0.10-0.64) and a higher behavioral intention score (r=0.378, 95% CI 0.11-0.59) were associated with an increased number of posts.

We found that 3 demographic characteristics were associated with the number of log-ins. Retired caregivers had an average of 27.4 log-ins as compared with 11.3 for nonretired caregivers. Both primary caregiver age (r=0.31, 95% CI 0.03-0.54) and the total behavioral intention score (r=0.38, 95% CI 0.11-0.59) were associated with more log-ins. We found that a higher SUS score was associated with an increased number of secondary caregiver invites sent (r=0.33, 95% CI 0.05-0.57).

This study establishes the acceptability and feasibility of CareVirtue use among care networks of individuals living with dementia. The results indicate that CareVirtue was perceived as highly usable and useful, with caregivers indicating a range of ways CareVirtue was useful to them, including practically, organizationally, and emotionally. We found that retirees, spouses/partners of the care recipient, and those who live with the care recipient were more likely to post more frequently in the journal. CareVirtue use was not correlated with caregiver age or education level.

Participants were confident in using CareVirtue, with confidence increasing over time, which aligns with their reported perceptions of CareVirtue as highly useful and usable. The qualitative analysis revealed that CareVirtue is useful across multiple dimensions including reducing burden associated with logistics and organization; providing emotional and social support; and facilitating documentation, tracking, and awareness across the care network.

Interestingly, participants’ perceived workload associated with their caregiving role increased over the study period. It is possible that this result was reflective of the burden of participating in the study. However, we also found that higher caregiving workload was associated with more frequent use of the CareVirtue journal. It is also possible that this consistent increase in perceived caregiving workload was related to the increased burden and isolation associated with the COVID-19 pandemic during which our study took place [34,35]. A recent survey study [33] found that caregiver burden was not associated with caregivers’ reported intention to adopt a mobile health (mHealth) intervention. Our finding based on the engagement with the technology intervention, that increased caregiver workload was associated with increased journal posts, provides additional insight into factors that may be influencing caregivers’ adoption and use of technology interventions. This finding also points to specific components, such as the ability to journal about daily experiences and emotions, that may be more useful during times of higher workload.

We found that usage varied along 2 axes: total usage and type of usage. There was heterogeneity in how the most engaged care networks interacted with the platform; for example, 2 care networks made heavy use of the calendar event feature (average of 141 events) with few posts (average of 41), while another care network made heavy use of the journal (257 posts) with only 1 calendar event. These results confirm the importance of technology interventions that can account for caregiver heterogeneity [36,37]—that caregivers are diverse individuals who have wide-ranging experiences, needs, and contexts. The need for interventions responsive to caregiver heterogeneity has been further supported by the identified importance of multiple component interventions that are tailorable to specific caregiver needs [5,9-11]. Our findings expand on this by providing insights into the acceptability and feasibility of a technology intervention at the care network level. Like studies focused on primary caregiver technology interventions, our results highlight the importance of designing flexible, multicomponent interventions that allow care networks to tailor their engagement according to their needs and what is most helpful to them. Importantly, this could also allow for tailoring over time, in which care networks can shift engagement as their needs change across disease progression.

This study used a sample size that is much larger than is typical for a feasibility study [38]. Doing so gave us the opportunity to leverage data analytics to provide insight into how care networks engaged with the platform and provided depth to our understanding of what components caregivers find useful [28]. Survey studies have been useful in capturing self-reported perceptions related to adoption and intention to use [39,40]. Findings such as ours along with others such as Øksnebjerg et al [41] can complement and expand upon findings from self-report studies by exploring caregivers’ engagement behaviors. Given caregiver and care network heterogeneity along with the evidence demonstrating the importance of flexible and tailorable multicomponent interventions [3,12-19], there is a need for future research to continue to explore caregiver and care network engagement behavior with technology interventions to provide additional insight and begin to build an evidence base regarding how to optimally tailor interventions and support engagement according to individual caregiver and care network needs. To do this, future feasibility trials could strive to engage larger sample sizes, enabling the use of artificial intelligence and machine learning to increase personalization. Further, efficacy/effectiveness trials that typically enroll larger samples could be used to explore engagement. Doing so may allow for an increased understanding of the relationship between engagement with the technology intervention and the health outcomes.

Although the purpose of this study was not to determine the effectiveness of CareVirtue on caregiver outcomes, previous research has demonstrated the potential of technology interventions to improve caregiver outcomes [3,12-19]. Research suggests that the significant unmet needs associated with support for communication and coordination among the care network may contribute to the often-suboptimal outcomes experienced by caregivers such as increased stress and burden [20,24,37]. Further, our findings related to caregiving workload and journal use combined with our qualitative findings provide some initial indication that caregivers may experience positive effects such as reduced burden, increased social support, and increased quality of life from using CareVirtue. Our immediate plan for future research is to conduct a randomized clinical trial to test the hypothesis that CareVirtue reduces caregiver stress and burden and increases caregiver quality of life.

Our results should be considered in light of certain limitations. First, although our sample size was much larger than is typical for a feasibility study [38], the sample size should be considered when interpreting the univariate results, as it is possible that a small group of people could be driving our findings. Second, although we achieved enrollment of diversity in terms of income and location within the United States, participants were primarily White, married, college-educated women of 60 years of age who lived with the care recipient. Future work will endeavor to achieve a more sociodemographically diverse sample in terms of race/ethnicity, education, age, and distance from the person living with dementia. Third, although this study found a broad range of care network sizes consistent with previous literature [20], it is likely that the context of the COVID-19 pandemic may have reduced the number of in-home supports, which may have influenced the number of care network members using CareVirtue.

The results of this study establish the acceptability and feasibility of CareVirtue use among care networks of people living with ADRD. This study also highlights the importance of designing flexible, multicomponent interventions that allow care networks to tailor their engagement according to their needs and what is most useful to them. The results of this feasibility study will be used to improve CareVirtue feasibility and acceptability in preparation for a subsequent randomized trial to test CareVirtue’s effectiveness in improving caregiver outcomes.