This study was a qualitative secondary analysis of semistructured interview data with participants of a mixed-methods study named MONITOR-OA [11]. The original study involved a 6-month proof-of-concept randomized controlled trial (RCT) to examine the efficacy of a technology-enabled counseling intervention for increasing MVPA among persons with knee OA and a qualitative component to describe the participants’ experiences of the intervention [11]. Fundamental qualitative description and conventional content analysis methods guided the qualitative component of the original study [26,27]. As a complete data set was already present, our secondary analysis used phenomenographic analytical methods and introduced a relational ethics lens to shape interpretations of the original interview data, in line with Varpio et al’s [28] description of a theory-informing inductive data analysis study design [29,30]. Relational ethics is a broad theoretical lens that continues to evolve from critiques of a strong individualistic perspective that dominated traditional bioethics discourse [31-34]. These critiques highlight the complex ways in which persons develop within (and are inherently shaped by) relationships (personal and institutional, past and present) that are an integral part of one’s life. They expand on traditional bioethics principles by locating ethical issues in the context of everyday relational settings [31]. Phenomenography is concerned with relations between a person and a specified aspect of the world as it appears to them [29]. Phenomenographic analytic methods and a relational ethics lens thus offer appropriate theoretical grounding to explore the particular focus of our study on everyday ethical issues experienced by participants in their relationships with themselves and the PT while using a physical activity wearable. Our approach rests on assumptions that reality is socially and experientially constructed, and to understand these realities, researchers need to explore the meanings constructed by individuals or groups.

The RCT component of the original study took place between November 2015 and June 2017 in Vancouver, Canada. Participants attended a 1.5-hour session, where they received (1) 15-minute group education about physical activity, (2) a Fitbit Flex (Google LLC), and (3) individual counseling with a study PT who was trained in motivational interviewing [35]. The individual counseling session followed the brief action planning approach, whereby PTs guided participants to identify activity goals, develop an action plan, and identify barriers and solutions [36].

Participants were asked to wear the Fitbit wristband 24 hours a day except during water-based activities or when recharging the battery. The physical activity data were wirelessly synchronized with Fitbit’s online Dashboard that could be viewed only by the participants and their study PT. During the intervention period, the PT reviewed the participants’ physical activity data on the Dashboard and reviewed their activity goals during 4 biweekly phone calls. Participants could also contact the PT via email in-between the scheduled calls. After the 8-week intervention concluded, all participants were invited to take part in an interview about their experiences.

The research protocol was approved by the University of British Columbia Behavioral Research Ethics Board (H14-01762) and was published in clinicaltrials.gov (NCT02315664). Informed consent to use interview data for the secondary analysis was obtained from all participants at the time of the original study.

Further details of the original RCT have been described elsewhere [11]. Briefly, individuals were eligible if they were adults living in Vancouver, Canada, with a physician-confirmed diagnosis of knee OA or who passed two criteria for early OA: (1) aged 50 years or older and (2) experienced knee pain during the previous year lasting more than 28 separate or consecutive days [37]. Individuals were excluded if they (1) had been diagnosed with inflammatory arthritis, connective tissue diseases, fibromyalgia, or gout; (2) used antirheumatic drugs or gout medications; (3) had previously undergone knee arthroplasty; (4) had suffered an acute knee injury in the past 6 months; (5) had a body mass index of 40kg/m² or higher; (6) had received a steroid injection or a hyaluronate injection in the last 6 months; and (7) were using medications which impaired physical activity tolerance (eg, beta-blockers), or had an inappropriate level of risk for increasing their physical activity. Participants were also excluded if they did not have access to a computer in their home or did not have a personal email address.

The original RCT had 61 participants, of which 56 completed the in-depth interview after the intervention. As this study aimed to examine the range of experiences among participants, sampling was mostly theoretical to maximize variation across demographic characteristics (eg, age, sex, and education). Our analysis focused on a purposive subsample of 21 of these interview participants (13 females and 8 males), ranging in age from 40 to 82 years (Table 1; participants chose their pseudonyms). Participants came from a variety of household compositions, with education levels ranging from high school graduate (4/21, 19%) to a bachelor’s degree or above (11/21, 52%), and annual household incomes ranging from under CAD $12,000 (US $9,750; 1/21, 5%) to over CAD $100,000 (US $81,240; 4/21, 19%).

The first author (JL) conducted 13 of the 21 in-depth one-on-one interviews, and the second author (GM) conducted the remaining 8 interviews. JL (cisgender woman) and GM (cisgender man) were PhD Candidates with approximately 10 years of experience each in qualitative interviewing. All interviews followed a semistructured interview guide including the following questions: Can you tell me about any physical activity you do? Since taking part in this study, have you noticed any changes in your physical activity? How easy or difficult has it been using the Fitbit? What about your experience with the study PT? Prompts and probes were also used as reminders during active listening, and considerable attention was given to the interviewee articulating their experience in their own words. Interviews (lasting approximately 60-90 mins) were audio-recorded and transcribed verbatim.

Phenomenographic analytical methods were used to describe the different (positive and negative) ways in which persons with knee OA experienced the physical activity counseling intervention and the various meanings they attributed to this phenomenon in their world [29]. Although the roots of phenomenographic research methods lie in learning studies, the methods have been applied to a variety of other issues inside and outside the field of education [38-41]. We use this analytic approach for two reasons. First, the MONITOR-OA RCT provided a complete data set of interview transcripts in which participants’ experiences using a wearable in a physical activity intervention was a prominent issue. Second, this study focused on exploring differences among the collective experience that persons with OA had when participating in the intervention. Both reasons aligned with phenomenography, justifying it as an appropriate approach for analysis.

We analyzed the data through a relational ethics lens, following Heaton’s [42,43] definition of supra-analysis, which is one type of secondary analysis involving examining pre-existing data from a new theoretical perspective, thereby transcending the aims and focus of the original research. We reused our self-collected data to examine it from a new theoretical perspective than the primary study, specifically a relational ethics perspective. Relational ethics places the principles of bioethics that are traditionally used to guide moral practice in health care (eg, autonomy, nonmaleficence, beneficence, and justice) within the context of close-up relationships [30,34,44]. It assumes that all relationships are moral and attends to the commitment to ethical action in one’s relationship to oneself and the other in every situation or encounter on an ordinary everyday basis [34,44]. Issues of relational autonomy, such as engagement and partnership, have been previously identified in perspectives of persons with arthritis on their use of wearable devices to self-monitor physical activity [20]. Building on this earlier research, our study focused on participants’ experiences of the benefits or drawbacks of using a wearable, with a particular interest in any impacts experienced in their relationships with themselves and with the study PT. Therefore, relational ethics is a suitable conceptual lens through which to continue the exploration of ethical issues identified from the perspectives of persons with arthritis.

Our analysis was informed by the 7 steps of phenomenographic analysis as described by Sjöström and Dahlgren [41]: familiarization, compilation, condensation, grouping, comparison, naming the categories, and contrastive comparison (Textbox 1). Data were reviewed carefully for themes that attend to morality in everyday relationships, including (but not limited to) self-control, engagement, and partnership [30,34,44]. Regular meetings were held with the research team, some of whom had experience in phenomenographic analysis, to enhance rigor and check, test, and probe preliminary findings [45].

Informed by calls to critically consider the relevance of notions of saturation within our study’s context, we did not consider saturation to be meaningful to our research objective and methodological orientation because we did not seek to saturate theoretical categories, themes, or data [46-49]. Our sampling and analysis ceased when the research team reached an agreement that findings were sufficiently varied to describe a range of participants’ experiences relevant to our research objective. Based on the previous experience of phenomenographers, 15 to 25 interviews are typically preferred in a phenomenographical study [50]. While we acknowledge that what is determined as an appropriate sample size for one qualitative study is not necessarily an appropriate sample size for another qualitative study [51], the combination of team agreement regarding content and the number of interviews compared to past phenomenographic studies supported our decision.

Phenomenographic analysis provided a rich interpretation of how persons with knee OA experienced benefits and drawbacks in using a Fitbit during their participation in a physical activity counseling intervention study. Our analysis revealed 3 main categories: (1) making choices about physical activity with or without a wearable, (2) emotional dimensions of adding awareness about physical activity, and (3) reviewing wearable data with the study PT: issues of accountability and trust. Key quotes are presented in Textboxes 2-4 to illustrate each category, and supplementary supporting quotes can be found in Multimedia Appendix 1.

Participants described how their Fitbit had influenced their choices to increase their walking to reach their daily step goal (Textbox 2, quotes 1-4). Some attributed human-like traits to their Fitbit when describing its influence on their activity choices (Textbox 2, quotes 1-3). For example, Hazel described her Fitbit as “a little person on my wrist…a little friend” that was a welcome source of friendly and gentle encouragement to meet her step goals. For her, Fitbit was “happy when I do 10,000 steps…just like a friend supporting me…There’s a gentle persuasion.” Others expressed how they felt pressured by their Fitbit to meet their activity goals. Martha, for example, felt ambivalently toward her Fitbit’s influence in her decision to walk more, reflecting that “it probably gives me some incentive to walk a little further just to placate the Fitbit…it nags you…I’m fine with that…it’s probably a good thing to have something that makes you get up and go.” Denny felt “forced to kind of do some more activities” by her Fitbit in the evenings when she felt too tired. Sansa indicated she felt a sense of responsibility to meet her activity goals and viewed the Fitbit as “there to keep me accountable.”

Some participants described certain days they did not engage with their Fitbit; for example, on days they were experiencing “too much pain” or on “some days I don’t care” to be more active. Zed found that Fitbit’s influence did “start me going” to meet step goals “in the beginning” of the study, and its influence “wanes over time” once his increase in walking became part of his habitual routine. Others explained that wearing their Fitbit did not add any value as they were making choices to meet their activity goals regardless (Textbox 2, quote 5). As a busy mother, Yoda found that Fitbit was generally irrelevant as meeting her activity goals was not her main priority (something for which she suggested “shows perhaps an inherent bad attitude”). She recounted, “I wasn’t invested that I absolutely had to do, come hell or high water, these steps so I’d be like yeah, I just didn’t walk very much today.”

Some participants highlighted there was an emotional dimension to the heightened awareness of their physical activity levels they experienced through using the wearable (Textbox 3, quotes 1-3). Some described how this added awareness about their activity levels prompted feelings of accomplishment and gratification, which fueled their motivation. For example, Daenerys recalled, “It lets me feel as though I’m accomplishing something every day…I feel pretty happy…and then it’s kind of fun to see how much more I can do.” Olivia also felt “instant gratification” as she could see improvement in Fitbit’s feedback on her step goals.

While Logan Kale felt “happiness, accomplishment” when she reached her activity goal most days, she also found that on days when she was “in too much pain” to reach her step goal, she would “try not to beat myself up over it” and “kind of get stuck on that hamster wheel of negative thoughts and have to zip it.” Biker commented that she did not “feel bad” when Fitbit data showed she had not met their step goal because “I like to exercise and so being active is not an issue for me…actually I kind of find it a kind of cozy feeling thinking ‘I haven’t completed everything. There’s still more to do.’” She expected, however, that others who were less active than her “might feel sad” if they did not meet a physical activity goal.

Participants described how their choices to be physically active had been influenced because the study PT had access to review wearable data on their physical activity (Textbox 4, quotes 1-2). For example, Denny described how she had “tried a little bit harder [laughs] maybe for a few weeks” to reach her activity goals because she knew the study PT had access to her wearable data and found this to be “very, very encouraging.” Darius also indicated that he felt a sense of accountability to meet the physical activity goals he had agreed with the study PT, as he recalled “a little bit of pressure…I need to watch to keep my promise.”

For one participant, accountability issues intertwined with trust issues in his relationship with the study PT when reviewing his wearable data together (Textbox 4, quotes 3-4). Recounting how his wearable data had not always accurately tracked his step count because it sometimes had “slipped into sleep mode,” Gavin described a phone call he had with the study PT in which he explained that his subjective account of his physical activity should be trusted as more credible than his wearable data, and he should not be held accountable, because “it wasn’t my fault I didn’t make my 10,000.” When he felt his Fitbit had tracked his step count more accurately, Gavin described this data as “some form of proof” that could be shared with the study PT “for her to gauge” his activity. He also commented that “if [Fitbit data] wasn’t there… then ‘oh yeah I climbed Mt. Everest this weekend’…I could’ve been making up anything about that,” emphasizing that, when accurate, Fitbit data may give a more reliable account of his physical activity than himself.

Our findings provide novel insight into different ways in which persons with knee OA experienced their use of a wearable positively or negatively during their research participation. Firstly, the contradictions in the data were fascinating. Participants experienced their wearable positively as a motivating influence and more negatively as a nagging reminder to be more active. From a relational ethics perspective, these findings shed light on how participants’ experience of their wearable impacts their autonomy positively or negatively.

Autonomy is a central notion in modern health care ethics that is understood as the capacity to direct one’s own choices freely and intentionally [31]. A relational approach builds on this traditional understanding of autonomy by focusing attention on relationships and interdependencies that may support or impair a person’s capacity for autonomy [52,53]. The relational autonomy perspective highlighted that many participants experienced the relationship with their wearable as a support to their autonomy. The device helped them take more control in making choices to be more active. However, some participants who described their capacity to make autonomous choices about their physical activity as impaired (eg, due to tiredness) experienced tension and ambivalence in the relationship with their wearable, as they described feeling pressured or forced to be more active at times when they did not entirely wish to move. These findings align well with previous research suggesting that using a wearable may be experienced as autonomy-enhancing or autonomy-undermining by persons with arthritis, regardless of whether they used the wearable to support physical activity during research participation or as part of their everyday self-management [20]. Leese et al [23] have reported some persons with arthritis experienced their wearable as a motivating or autonomy-enhancing support in their everyday self-management when used as a “nice reinforcement” to an already physically active lifestyle.

Findings in our study also indicated positive and negative experiences of participants using a wearable, depending on how freely they were able to direct their own choices to be active given their specific situation or set of circumstances. They resonate with a relational ethics approach that recognizes exercising autonomy requires relevant capabilities, which are dynamically shaped by a person’s situation or set of circumstances [54]. They also resonate with calls from some health professionals for academic literature to facilitate the positive development of self-tracking technology in self-management by reflecting on context-relativity (rather than focusing on “ideal” situations) [55]. We posit that further research is thus warranted to build a greater understanding of the everyday contexts in which the use of a wearable may be experienced as autonomy-enhancing or autonomy-undermining by persons self-managing chronic illness.

At this early stage of the potential integration of wearables into arthritis self-management, our findings can contribute to ongoing conversations in clinical practice. They suggest health professionals may wish to carefully consider a person’s capability to make autonomous choices about their physical activity if using a wearable in their everyday self-management. They, therefore, align well with research suggesting how health professionals may tailor their support for arthritis self-management in ways that take a person’s capacities to engage in physical activity within a 24-hour day context into account [56,57]. Further research could guide forms of support that health professionals may offer to persons with arthritis using a wearable who struggle to freely direct their own choices about physical activity in their everyday self-management. Without this support, using a wearable in everyday self-management may be experienced as a nagging or autonomy-undermining influence by some persons with arthritis, adding to a struggle to feel in control of their choices to be more active. Evidence exists to indicate that feeling this sense of control in self-management is at least one of the mechanisms responsible for improvements in health behaviors and health status [58-60].

Secondly, our findings build on previous research indicating that persons with arthritis experienced an enhanced awareness of their activity levels when using wearable technology [25,61]. The insights emphasized how a heightened awareness impacted participants’ emotions positively and negatively. Some participants experienced feelings of accomplishment when their wearable data illustrated that they had reached their physical activity goal, which often fueled their motivation to do more. One participant, however, experienced negative thoughts on days that his wearable data indicated he had not reached his physical activity goal. Mercer et al [25] also found that persons with chronic illness (including arthritis) experienced negative feelings in using a wearable during a 15-day research period as they were concerned they were not sufficiently active. From a relational ethics perspective, these experiences speak to a theme of embodiment, emphasizing the importance of complex emotions or feelings in a commitment to ethical action in one’s relationship with oneself and others [34,62]. It remains unknown how others with arthritis may be emotionally impacted if using a wearable in a “real-world” context, outside of research participation. Our findings, therefore, raise questions about how the use of a wearable device within a counseling program may impact the emotional wellbeing of individuals, particularly as they evaluate their capacity and progress in managing their health.

Thirdly, findings indicate how participants experienced issues of accountability and trust differently when reviewing their wearable data with the study PT. Some participants were motivated by a sense of accountability to “keep my promise” to meet the activity goal agreed with the study PT. For some of these participants, wearable data served as “some form of proof” that the study PT could trust to “gauge” whether they had met this goal. One participant, however, expressed how his wearable data was not to be trusted at times and recalled reaching a mutually satisfactory interpretation of this data as inaccurate with the study PT. These different experiences illustrate how sharing wearable data helped build mutual trust and engagement between participants and the study PT. At the same time, this finding also indicates how sharing wearable data threatened to undermine mutual trust and engagement within this relationship.

The fundamental role of building trust in interpersonal relationships in health care has been emphasized elsewhere through a relational ethics lens [62-64]. A relational ethics lens can also be explored here through the relational theme of engagement. Genuine engagement is understood by Bergum [34] to be “located in the shared moment when people have found a way to look at something together, freely accepting or declining the interpretation that each other offers, until they reach a meaning they both affirm.” Our findings support the general opinion previously expressed by persons with OA that sharing their wearable data may enhance their communication with health professionals in everyday self-management under specific conditions (eg, if there was a good rapport already established in the relationship or if a health professional would welcome the wearable data being shared) [20]. They also raise questions about the sense of accountability experienced when sharing wearable data with a health professional, in terms of how far this may be a burden to persons with arthritis in their everyday self-management [65,66]. Further research is therefore needed to gain a better understanding of the relational conditions in which persons with arthritis may experience issues of accountability, trust, and engagement positively or negatively when sharing wearable data with their health professionals in everyday self-management.

There were limitations to this study. As a secondary analysis, the data were not created with the relational ethics lens in mind, and therefore potentially important experiences might not be fully elicited in the interviews. Nonetheless, transcripts were purposively selected to offer sufficient variation in participants’ experiences relevant to our objective. A phenomenographic approach also allowed us to identify overarching meanings that crossed transcripts and were implicitly presented by the collective group. To better examine the transferability of findings, further research is needed to explore the experiences of a more diverse sample using a wearable in the context of their everyday self-management of arthritis outside of research participation. It may be that persons of diverse genders or cultural backgrounds, for example, encounter different experiences. Our subsample, however, is varied and represents a typical OA group in terms of age and sex.

Our findings provide insight into different ways in which persons with OA experienced their use of a wearable during participation in a physical activity counseling intervention study positively or negatively. Drawing on a relational ethics lens, we identified how issues of relational autonomy, embodiment, accountability, trust, and genuine engagement were present in these experiences. These issues have implications for learning how to develop and implement wearable-enabled physical activity programs to support arthritis self-management in ways that seriously factor in ethical considerations. We present these salient ethical issues for further discussion and to guide future empirical investigation of the use of wearables in arthritis self-management.