Wearable health monitoring devices have seen a rapid rise in capability and popularity over the last two decades. These small wireless devices can monitor movements, improve physical activity, and facilitate ageing-in-place. Wearable devices temporarily and noninvasively attach to a person without hindering their movement and are often intended to be worn continuously. Examples include activity trackers (eg, Fitbit and smartwatches), fall detection devices, electromyography patches, and smart clothing.

Although older adults are not core consumers of wearable devices, their use of digital health technologies is increasing [1] in tandem with the expanding technological capabilities of wearable devices. Wearable devices can support “active ageing,” the process of enhancing quality of life as people age [2]. Technology creates an enabling environment that restores function and expands the participation of older adults in their health. Remote monitoring using wearable devices can aid independence and encourage older adults to manage stable chronic conditions by themselves. Clinicians can track patients’ health status remotely and communicate via video-based consultations [3]. Current wearable devices possess the ability to monitor a number of health metrics, including heart rate, blood oxygen levels, body temperature, physical activity, sleep, and blood pressure [4]. The older adult population is vulnerable to changes in their health conditions and may be burdened by frequent clinical visits. Wearable devices are well suited for monitoring older adults because they convey up-to-date health information and track health metrics over time. Wearable devices are intended to be worn continuously. For example, fall detection devices are worn all day, as falls occur unexpectedly. As these devices are used frequently, it is important to understand the barriers to acceptance and adherence. Factors such as trust, functionality, added value, ease of use, cost, stigma, and fear of dependence are examples of barriers to adoption [5].

Researchers have used a variety of methods to collect information from older adults regarding the acceptability of wearable devices: surveys [6,7], wear time [1], diaries [8,9], interviews [9], and focus groups [10,11]. Some studies collected information about general preferences regarding device design [11]; others allowed participants to interact with several devices before asking about preferred design features [12], in which participants used a wearable device for multiple days and then provided feedback [13-15].

Qualitative research methods are well suited to examine the user experience and may offer explanations for unexpected or anomalous findings in quantitative data [16] or uncover usability barriers that quantitative approaches often miss. Systematic reviews that combine the findings of multiple qualitative studies can identify common factors among studies and generalize their findings. No qualitative systematic review exists on older adults’ experiences of using any form of wearable device. Although each user experience is unique, a synthesis of studies may lead to a richer understanding of the integration of devices into the lives of users. Our objective is to better understand these experiences to inform future research efforts and to inspire device design to ensure a successful user experience.

We aim to apply a qualitative meta-synthesis process to the available qualitative data on older adults’ experiences with using wearable devices. Meta-synthesis is a form of interpretive synthesis that can be used in the review and evaluation of qualitative research studies. Our meta-synthesis is based on the principles of meta-ethnography [17], a method designed by Noblit and Hare [17] to synthesize ethnographic studies. A meta-synthesis differs from traditional meta-ethnography in that it allows for a variety of data analysis techniques besides ethnography (eg, phenomenology and grounded theory) to be synthesized together. It is an inductive method that compares, translates, and integrates concepts across studies, while also preserving the context of the primary data. This meta-synthesis is more in-depth than previous systematic reviews on wearable devices that summarized a group of studies [4,5,18-23], analyzed a series of trials [24], or reviewed the state of the art [25,26]. Our overall goal is theory development, higher-level abstraction, and generalizability for making this group of qualitative findings more accessible [27].

Our overarching research question is “What is the experience of older adults who took part in multi-day wearable device trials and what factors contribute to acceptance and use?”

The inclusion and exclusion criteria (Textbox 1) were designed to accommodate various study designs as long as they contained the following core structure: an older person using a wearable device for multiple days and the subsequent qualitative analysis of that participant’s experience in relation to the wearable device. We created an inclusive search strategy to locate studies that used unusual jargons or unconventional study designs (Textbox 2). We also reviewed the search terms used in other systematic reviews in this area, consulted with colleagues in clinical and technical expertise in the area, and piloted various combinations of search terms to assess the sensitivity of the terms. KM searched four databases—CINAHL, APA PsycINFO, PubMed, and Embase—from January 2015 to January 2020 for studies published in English. Our date range intentionally excluded older wearable devices to minimize the differences between the capabilities of the devices used in the studies. Insights from before this period were recorded in a systematic review by Bergmann and McGregor [28]. This search was supplemented by searching Google Scholar, forward and backward searches of citation lists, and the publication lists of prominent researchers in the field.

KM exported the search results, removed duplicates, screened all titles and abstracts for inclusion, and reviewed eligible full-text articles against the inclusion or exclusion criteria (Textbox 1). LK screened a random sample of 200 abstracts (and full texts where the abstract indicated potential for inclusion) and confirmed the consistent application of the inclusion and exclusion criteria. A senior researcher, ST, provided guidance when eligibility based on a full-text review was unclear. Reasons for exclusion were recorded for all excluded studies (Figure 1).

Data items extracted included information about the publication (date, authors, and study aims), study process (design, methods, and analysis), participant characteristics, device types and features, trial length, and relevant primary qualitative data (themes and quotations).

Although not required in a meta-ethnography, we assessed study quality to facilitate the critical reading of each study to gauge its potential contribution to the analysis (see Table S2 in Multimedia Appendix 1 [9,13-15,29-44] for the checklist). We used the Evaluation Tool for Qualitative Studies (ETQS) [45], as this provides detailed instructions on applying the evaluation criteria, unlike the CASP (Critical Appraisals Skills Programme) tool [46,47]. The ETQS guides the appraisal of the phenomenon studied and context issues; ethics; data collection, analysis, and researcher bias; and policy and practice implications [45]. Two authors (KM and LK) independently conducted the quality assessment. No studies were excluded based on ETQS results, as this often reflects the level of reporting transparency, rather than the actual research processes used [16,48].

The analysis (Textbox 3) was guided by a meta-ethnographic approach [17,48,49]. Initially, papers were read and reread to familiarize researchers with the study context, design, and findings. Individual themes were extracted and recorded using separate index cards. As the studies were methodologically heterogeneous, we preserved the authors’ original themes (and wording), where possible. Where appropriate, we extracted additional themes from the “discussion” and “conclusion” sections. Where studies were highly descriptive or simply listed participants’ quotes, we coded the “results” and “discussion” sections and generated themes from the presented data, noting that specific quotes, without the full conversation context, were challenging to code. Studies that presented minimal or overly descriptive results were mainly used to support or refute the themes identified in high-quality studies.

The themed index cards were physically juxtaposed and grouped into categories based on patterns of meaning, as related to the research question. Categories and subcategories were refined iteratively through constant comparison within and across studies. Each category was compared against each original study using (1) reciprocal translation (recognizing reoccurring themes or concepts across studies) and (2) refutational translation (recognizing dissimilar themes or concepts across studies, not explained by contextual factors). When all the data were collated and interpreted, several key concepts were defined (third order) and synthesized to develop an integrative “line-of-argument.”

We tracked the preferred and disliked device features throughout the analysis process. Where relevant, we used specific device features to support the key concepts. We summarized the preferred and disliked features, but no frequency analysis was performed because each study used different devices and not all studies included participant feedback on device features.

We reported our results in line with the eMERGe guidance, which has been described for use by researchers conducting meta-ethnography [50]. The search strategy results are presented in a Preferred Reporting Items for Systematic Reviews and Meta-analyses flow diagram (Figure 1).

The database search returned 1971 results (Figure 1). After title or abstract screening and duplicate removal, 51 full-text records were reviewed, and 14 were considered eligible for inclusion. Backward or forward searching uncovered 6 additional eligible records to reach a total of 20 records for the synthesis.

Of the 20 included records, 2 evaluated fall detection devices [14,29], 1 tested an ankle-worn step counter [30], and the remaining 17 examined wrist-worn activity trackers. The duration of use ranged from 3 days to 24 months. In some studies, users completed multiple trials with different devices [9,31-33]. In others, participants were randomly assigned to one of several devices [34,35]. The views of 349 participants (age: range 51-94 years) were synthesized, including those with previous breast cancer [32], obesity [36,37], resolving heart failure [30,37], Parkinson disease [38], dementia [15,39], and walking aids [9,29] and those who were fully independent and healthy [9,13,14,29,31-35,37,40-44]. Table 1 summarizes the results of the data extraction process.

This study found four key concepts, comprising 12 subthemes that characterize the collective experience of trial participants (Textbox 4).

The key concepts (order not indicative of prominence or salience) are (1) openness to engage and functional ability of the user; (2) motivation for device use; (3) integration into daily life; and (4) device features.

First-order quotations (raw, primary data, ie, direct participant quotations) and second-order (authors’ interpretations of their primary data) interpretations were used to support the analysis of the Translation section. Throughout the Results section, first-order quotations (primary study participants) are indicated in italicized quotations and second-order interpretations (primary study authors) are indicated in italics.

This is the first study to systematically review and synthesize the qualitative literature on older adults’ experiences with wearable devices. This meta-synthesis collated the experiences of 349 trial participants and presented the key factors that influence user acceptance and adherence. These factors include intrinsic and extrinsic motivation to use the device; the purpose of the device and how it relates to the user’s expectations and needs; and the ease of use and functional ability of the user. The user’s appraisal of these factors determines the level of value added by the device to the life of each user.

Motivation for device use comes in two forms (intrinsic and extrinsic) and encompasses many aspects of the user experience. According to our line-of-argument synthesis, motivation influences both device adoption and added value. Motivation seems to be as, if not more, important for older adults than the actual device features. Moreover, the user’s needs and the support structure around the device—aspects that are often overlooked—seem to play a crucial role in long-term adoption.

Our bottom-up inductive qualitative synthesis supports the findings of existing theory-bound models of technology acceptance. It was not intended from the outset that our conceptual model would tie in with quantitative models such as the Technology Acceptance Model (TAM) and the value-based adoption model (VAM). We felt that the TAM and VAM could be used to structure and contextualize our findings. These models use quantitative surveys to test hypotheses about factors that predict the intention to use. For example, the TAM shows that perceived ease of use, perceived usefulness, and attitude toward the system predict intention to use [51]. The development of these models does not involve trial components or qualitative methods. Therefore, our study should not be directly compared with the TAM or any other acceptance model and should instead provide inspiration for hypotheses to test in future iterations of wearable device acceptance models.

Originally designed to describe the acceptance of information services in organizations [51], the TAM has only recently been applied to wearable devices [52-55]. In contrast to the TAM, the authors of the VAM recognized that most consumers adopt mobile technologies for personal purposes and that the cost of voluntary adoption is borne by the individual, not the organization [56]. According to the VAM, perceived sacrifices (cost and technicality) seem to have a greater impact than perceived benefits (usefulness and enjoyment) on perceived value. Many participants in our review alluded to this balance between sacrifices and benefits. They described how a device that disrupts their routine or limits their independence is not worth the hassle, especially if the features (eg, counting steps) are not beneficial to them.

The TAM and VAM inspired the recent development of a smart wearables acceptance model for older adults by Li et al [57]. Along with established acceptance factors such as perceived usefulness [51], Li et al [57] included older adult–specific factors such as self-reported health conditions, perceived social risk, performance risk, and social influence. Their results supported their hypothesis that facilitating conditions positively predicted intention to use, a finding supported by our study. They also showed that the self-reported health status is a negative predictor of use, suggesting that older adults with a better health status are not likely to require these technologies. Our results show a similar trend; older adults are motivated to use a device if they have a need (eg, monitor symptoms and fall risk). In contrast to our findings, 95.9% (140/146) of their participants perceived minimal or no social risk when using a wearable device. This may be because their participants were not offered the opportunity to wear the device in public. Several of the participants in our review (who did wear their devices in public) described feelings of embarrassment or stigma when the device intrudes on the user’s life.

Our review points to age-related factors that can influence acceptance, such as experience with technology and openness to engage. A systematic review of factors influencing acceptance of technology for ageing-in-place found a similar phenomenon and discussed the effect of age and chronic illness on the acceptance of vital sign monitoring systems [58,59]. The systematic review also highlighted the impact of social support from family, friends, professional caregivers, and peers [58]. In their 2014 review of determinants and barriers, Lee and Coughlin [60] described eight similar factors (value, usability, technical support, social support, emotions, independence, experience, and confidence) and two additional factors that our review did not uncover (affordability and accessibility) [60]. Similar findings from these studies indicate a convergence of the field toward an understanding of the key factors that influence adoption and adherence.

When designing future wearable device acceptance models for older adults, researchers should consider the multiple stages of device use that follow the initial “intention to use.” Furthermore, in the user experience, concepts such as added value become relevant, which may have a different set of predictors than the initial intention to use. In the development of the Senior Technology Acceptance and Adoption Model (STAM), Renaud and van Biljon [61] related certain acceptance factors to adoption stages. The STAM includes factors such as confirmed usefulness and ease of learning and use. Using qualitative methods to develop their model, Renaud and van Biljon [61] used these factors to explain why older adults do not reach the final adoption phase and never fully accept technology. Yu-Huei [62] adapted the STAM to wearable devices and added two additional factors, information source and group behavior, which emerged from their qualitative analysis of older adults in Taiwan.

Our study proposed several predictors that could be tested as a part of future model development studies. First, motivation is key and seems to be a constant driving force throughout the user experience. Researchers should question users on both intrinsic and extrinsic motivators to see if these factors predict integration. Second, the purpose of the device (and whether the user aligns with that purpose) should be investigated as a predictor of device adoption and added value. Finally, ease of use should be considered within the context of older adults, as done by Yu-Huei [62]. The technical abilities of older adults may differ from those of younger generations, and certain medical conditions may hinder functional abilities.

Researchers play a role in validating (or refuting) the findings of this review. Older adults require a specific wearable device acceptance model because they are a distinct population from the individuals used to develop the TAM and other wearable device acceptance models [54]. As opposed to smartphones and embodied conversational agents, wearable devices are designed to be used individually and continuously, which leads to a unique set of influencing factors.

For clinical trial researchers, we stress the need to provide extrinsic motivation for their participants by conveying the importance of the device. They should also provide training and technical support to facilitate ease of use. Clinicians using wearable devices in their private practice can provide extrinsic motivation by clearly explaining the device’s purpose and the meaning of measurements to their patients. They should also provide encouragement and technical support. They can support intrinsic motivation by learning about their patients’ health goals and desires to use a wearable device.

Although not explicitly stated, several of the included studies used a user-centered design approach [9,14,29]. A user-centered design, though often overlooked, is essential in the development of wearable devices, particularly if the designers are not a part of the intended demographic. Designers, clinicians, and researchers can collaborate with older adults to address all aspects of wearable device use. As the studies in our review demonstrate, trials and qualitative data are valuable tools for designers using the user-centered design approach to explore the long-term use of their products. Some device features become relevant only after a period of usage. For the benefit of designers, we summarized some of the features that were commonly discussed in our review. The study participants discussed many design features that are specific to the devices in question, and thus features could not be compared between the studies. However, we can make a few generalizations. First, designers should not take for granted that older adults will accept every design feature. Second, a device should serve the purpose, and the primary function of the device should be reliable and easy to use; some older adults are interested in advanced features, but not all are. Third, participants disliked unnecessary interactions with the device, such as frequent charging, responding to alerts, and maintaining a Bluetooth connection to their smartphone. An ideal device requires little maintenance and only requires interaction to monitor the data and obtain device feedback. Finally, a wearable device should be easy to take on and off, comfortable to wear at night, and waterproof. It should cause minimal disruption in users’ lives, be aesthetically pleasing, and should not draw attention to the user or single them out as a patient; it should not interfere with clothing; and it should have a silent mode to prevent unnecessary disruptions.

Measures such as practicing reflexivity and using 2 reviewers maximized the quality of this meta-synthesis. The authors are an experienced multidisciplinary team (geriatric medicine, psychology, epidemiology, and engineering) with expertise in qualitative approaches. By following the eMERGe reporting guidance (see Table S1 in Multimedia Appendix 2 [50]), we communicated our methodology with transparency, including our study’s limitations. Although both reviewers (KM and LK) collaborated to generate the search strategy and inclusion criteria, because the second reviewer only screened a sample of the references, some relevant studies may have been excluded during the screening process. Our database search only included 70% (14/20) of the included studies. The additional 6 studies were located by manually searching the reference lists and searching Google Scholar, highlighting the limitations of our search strategy. For example, we should have included the search term “senior*” to find studies such as the Abouzahra and Ghasemaghaei [13] study. It is worth noting that this issue is more common in qualitative reviews than in quantitative reviews because of the poor and inconsistent indexing of qualitative research in databases [63]. While we searched four databases, searching for additional databases (eg, Scopus and ISI Web of Science) would have strengthened our study.

The content of our results and the line-of-argument conceptual model were contingent on the data collected from the broad inclusion criteria. This is both a strength and a weakness of qualitative syntheses; it affords reviewers the flexibility to uncover new ideas but it dictates which questions can be answered. Some studies lacked rich descriptions and interpretations of their findings or did not provide sufficient context (about the sample, the device, or the procedures). This limited the contributions of some studies to the meta-synthesis [64], regardless of their ETQS quality score. Studies with low ETQS scores often fell short of simple aspects, such as not reporting the location of the study. Ultimately, the results of this study are based on the synthesis of qualitative data, which is inherently subjective. Our line-of-argument and conceptual model provide suggestions, but the full development of a wearable device acceptance model for older adults will take place with a more rigorous study design.

All studies included in this review were published in English and conducted in Western countries. The findings may not represent countries with different cultures, access to wearables, and income levels. Several studies included short trial periods and a few participants. Each study evaluated a different device (or set of devices), which restricted comparisons between studies. Future research would benefit from long-term trials using in-depth qualitative methods to evaluate the drivers of acceptance and adherence. Future research must also include the views of older adults who use wearable devices as part of clinical care, not just a research trial. An alternative set of predictors might be relevant to participants who use a device for a specific health purpose.

This review found that several key factors influence the acceptance and use of wearable devices by older adults. These include intrinsic and extrinsic motivation for device use, ease of use, device purpose, and perceived added value to the user’s life. Designers, clinicians, and researchers should be aware that useful device features alone do not lead to continued use. To overcome the usability barriers (eg, limited technical ability), an older adult must be motivated to use a device because it serves a useful purpose. A support structure should be placed around the user that fosters motivation, encourages engagement with peers, and adapts to the user’s preferences. Future research should evaluate our conceptual model by validating our proposed predictors and conducting long-term wearable device trials that use qualitative methods to comprehensively address the multiple stages of device use and the many factors that contribute to adherence.