The smart health intervention was a 4-month, single-arm trial designed to develop an in-home, technology-based intervention and assess its feasibility for caregivers of people living with ADRD [16,17]. Caregivers received this study’s devices (ie, a laptop, a microphone connected to the computer, a router, and a smartphone). They placed them in an area of their home where they frequently interacted with their care recipients. We captured acoustic signals from the caregiving dyads through the microphone and used a deep machine-learning approach to identify stressful situations [16]. The system then sent a real-time stress management tip to the caregivers’ study smartphone via the ecological momentary assessment (EMA) system [16,17]. Examples of tips sent to caregivers included practicing body scan and deep-breathing exercises, taking a time out, or engaging in enjoyable activities. The messages encouraged caregivers to follow the tip and to report its usefulness. Caregivers were also asked to answer daily and weekly questions about their physical and emotional health status and the overall usefulness of the stress management tips [17]. We provided US $50 compensation to each caregiver and care recipient three times—1 month after starting the intervention, halfway through, and at the end of the intervention.

Recruitment took place between January 2021 and September 2023 using convenience sampling. Two main recruitment methods were used to recruit eligible participants: (1) on-site engagement at a memory clinic within a Midwestern academic medical center and (2) advertisements on social media platforms. At the memory clinic, a recruiter distributed study brochures to individuals who visited the clinic and might be eligible for this intervention. Once the caregivers expressed interest in this study, the recruiter shared their contact information with this study’s personnel (EK and KMR). This study’s personnel contacted the caregivers to explain this study’s procedures in detail and confirm their eligibility.

We also advertised the smart health intervention via social media platforms, such as Facebook (Meta) and Instagram (Instagram from Meta), to recruit caregivers across the United States. We allocated between US $100 and US $300 to recruit participants through social media platforms for 1 week of recruitment postings. The uploaded posts included this study’s title, primary text, target age and gender, study locations, and a link to a website containing this study’s detailed information. Individuals who saw this study’s advertisements and were interested in this intervention contacted the principal investigator (KMR) by phone or email. In addition to these two methods, individuals also contacted the principal investigator after finding the information about our study on websites listing clinical trials, such as ClinicalTrials.gov [18].

We recruited participants who met the following inclusion criteria: (1) aged 21 years or older, (2) residing with and caring for older adults living with ADRD, (3) not receiving payment for providing care, (4) having well-functioning home Wi-Fi, (5) receiving a score above 3 on the Revised Memory and Behavior Problems Checklist, (6) being fluent in English, and (7) agreeing to participate [17]. During the screening call, those who met the eligibility criteria provided verbal consent and signed the electronic consent form. Given that this intervention involved monitoring acoustic signals while caregivers communicated with their care recipients, care recipients were also required to sign the consent form. If care recipients could not consent, their caregivers provided proxy consent.

The smart health intervention was approved by the institutional review board at this study’s institution (2019B0406) and was registered with ClinicalTrials.gov (NCT04536701) on September 3, 2020, and completed on December 31, 2023. All participants provided written, informed consent before the data collection. Participants were informed about this study’s purpose, duration, location of data storage, potential benefits and risks of involvement in the intervention, and contact information for the principal investigators.

We used SPSS (version 29; IBM Corp) for descriptive analyses, including frequencies, percentages, means, and SDs, to present the characteristics of recruitment and participant demographics. We also used a note-taking approach during discussions with a recruiter and team members to capture key points on the topic.

Figure 1 presents the recruitment flowchart and reasons for nonparticipation. We attempted to screen 201 caregivers through social media (125 individuals), memory clinic (63 individuals), and other sources, such as ClinicalTrials.gov or unknown routes (13 individuals). Of these, 22 caregivers and their care recipients agreed to consent; only half completed the baseline questionnaire and began the intervention. Six of the 11 participants were recruited through social media, and the remaining 5 were recruited through on-site engagement at the memory care clinic (Figure 1).

Among caregivers excluded, 82 did not respond to the screening contact, 30 declined to participate due to privacy concerns, 29 were not interested in the intervention, and 26 considered this study’s protocol to be burdensome. Eleven respondents failed to meet the eligibility criteria, mainly because they were not primary or live-in caregivers, their care recipient was hospitalized or deceased, or they lacked functional Wi-Fi at home. Three caregivers expressed interest but later decided not to participate after further consideration. Other reasons for nonparticipation included (1) not providing written consent on a consent form after providing verbal agreement, (2) frequent travel plans, (3) having multiple household members besides the caregiving dyad, (4) planning to institutionalize their care recipient soon, and (5) providing incorrect contact information. Further, 11 caregivers who consented ultimately withdrew from this study for reasons such as no longer wanting to participate, failure to reach out, and the death of the care recipient.

Table 1 presents the demographic data for the participating caregiving dyads. The average ages of caregivers and care recipients were 60.09 (SD 13.52) years and 76.73 (SD 8.04) years, respectively. Most caregivers were female (n=8, 73%), while most care recipients were male (n=7, 64%). Most caregiving dyads were non-Hispanic White (n=7, 82% of caregivers; n=9, 91% of care recipients). Education levels varied, but all caregivers had attained education beyond a high school degree. Regarding employment status, 8 (73%) caregivers and all care recipients were either unemployed or retired. Eight (73%) care recipients and 7 (64%) caregivers were married, with caregivers typically being spouses. The average amount of caregiving hours per day was 21.18 (SD 6.37) hours, and the average caregiving duration was 34.82 (SD 20.26) months.

This study aimed to identify recruitment challenges in a smart health technology intervention for caregivers of people living with dementia. We observed a low recruitment rate and identified various reasons for nonparticipation among caregivers. Beyond nonresponsiveness to screening, the most common reasons for nonparticipation included privacy concerns, a lack of interest, and perceptions of this study as being burdensome. The demographic characteristics of participating caregivers showed that the majority were mainly female, White, spousal caregivers, and retired. Recruitment challenges were observed both on-site and through social media. On-site engagement challenges included the transition to remote medical appointments and staffing issues. At the same time, social media-based recruitment faced limitations related to the context of postings, budget considerations, and balancing demand and supply.

We observed a lower recruitment rate through social media (6/125 individuals) than on-site engagement (5/63 individuals). Previous literature has reported mixed results regarding the effectiveness of using social media for recruitment in intervention research [13,20-26]. While some studies have suggested that social media recruitment could be costly and ineffective [20-22,24], others have argued that social media could benefit the recruitment of potential target participants [13,23,25,26]. These conflicting findings highlight a need for further research to explore differences in recruitment rates across recruitment channels in technology-based interventions for people living with ADRD and their caregivers.

Caregivers of people living with ADRD may be anxious about sharing their data to use technology [27-29]. Additionally, they often prioritize dementia care and pay more attention to clinical trials focused on treating a specific disease than trials targeting their health [30]. While multiple pharmacological clinical trials may be ongoing in a clinic, caregivers of people living with ADRD may overlook nonpharmacological intervention studies. Moreover, they already spend considerable time and energy caring for their care recipients, and they may perceive participating in an intervention study, especially a technology-based trial, to be overly demanding [29,31,32]. Therefore, intervention research using technology should use strategies that reduce the perceived burden, such as providing training, using user-friendly platforms, or offering nondigital alternatives, including daily logs for those uncomfortable with technology in the caregiving population [8]. These tailored approaches have the potential to enhance the enrollment in dementia caregiving research.

Being female and non-Hispanic White are common characteristics of caregivers of people living with dementia in the United States [1], and spousal dementia caregivers are more likely to be retired [33]. Compared to male caregivers, females are more likely to sacrifice personal time, which can lead to a more significant burden and depressive symptoms [1]. Spousal dementia caregivers are also at a higher risk for depression, with rates being 2.5 times higher than that for nonspousal caregivers [1]. Caregivers with these characteristics may be more willing to participate in caregiving research on stress management. Still, the lack of demographic data on nonparticipants makes it difficult to confirm this. Therefore, future studies should gather information from participants and nonparticipants to better understand why some caregivers choose not to engage in research.

Despite our efforts, the strategies implemented in the smart health intervention needed to address recruitment barriers effectively. First, we needed more complete data on the socioeconomic status and race or ethnicity of potential participants who did not enroll in our study. Research shows that dementia caregiving dyads from lower socioeconomic backgrounds are less likely to participate in studies involving technology due to unfamiliarity and limited access to technological devices [31,34]. This gap in our data limited our ability to access the potential impact of the backgrounds of caregiving dyads on the recruitment rate.

Second, concerns about this study being too demanding persisted throughout recruitment, likely due to caregivers’ unfamiliarity with the technology. This study’s design, which included real-time messaging through the EMA system, may have contributed to this perception. However, an intervention study using the EMA system reported that the EMA approach did not bother participating caregivers [35]. Nonetheless, more research is needed to understand how the EMA system affects caregivers’ daily lives while capturing data [36].

We could not manage the recruitment challenges, such as the transition to remote medical appointments and staffing issues, through on-site engagement. Recruitment might have improved with more advertising before or after virtual medical appointments, greater research personnel involvement, and additional resources; however, barriers to participating in telehealth visits made implementing these strategies unfeasible. The rise in health care since the COVID-19 pandemic—an approximately 150% increase in telehealth visits [37]—posed a barrier to recruiting caregivers in person. Future studies should explore ways to overcome these challenges when recruiting participants with limited in-person contact.

Despite these limitations, the findings of this study can inform strategies for addressing recruitment challenges in dementia caregiving research. Our findings offer insights into the reasons for nonparticipation and how to address participants’ concerns regarding the smart health intervention. The findings may help researchers develop better strategies to improve recruitment rates in similar studies, offering caregivers opportunities to enhance their health and the quality of care they provide to their loved ones.