Adherence to medication for chronic conditions is a complex issue that is sometimes addressed by digital, web-based, mobile, and telehealth interventions [1-3]. We developed a new intervention involving a web-based adherence monitoring system with health professional support to help adults with cystic fibrosis (CF) increase their adherence to nebulizer treatments. Although not promoted as telehealth, the intervention has key components of web-based monitoring and web-based behavior change materials for use with the chronic condition of CF. CF is a genetic life-limiting condition affecting approximately 100,000 people worldwide [4] in which mucus builds up in the lungs, digestive system, and other organs, leading to difficulty in breathing, respiratory infections, and ultimately death. Patients with CF need to take preventative treatments such as antibiotics and mucolytics through nebulizers and airway clearance, often alongside many other treatments creating a complex, onerous treatment regime [5,6]. Adherence to nebulizer treatments is often low, with objectively measured adherence found to average 36% compared with 80% estimated by patients [7]. Low adherence to nebulizer treatments has been linked to worse health outcomes, including decreased lung function, increased pulmonary exacerbations requiring treatment with intravenous antibiotics, and higher service costs [8-11].

We tested the intervention in a feasibility study that included a pilot randomized controlled trial (RCT) and a mixed methods process evaluation. We found that a full-scale RCT was feasible with numbers of pulmonary exacerbations as the primary outcome. The pilot RCT identified that mean changes to adherence, a key secondary outcome, were 10% higher in the intervention arm (95% CI −5.2 to 25.2) [12]. The mixed methods process evaluation found that the intervention was feasible and acceptable to patients, interventionists, and health care professionals in the multidisciplinary teams (MDTs) offering wider care to patients with CF. As part of the process evaluation, we conducted qualitative interviews with patients and interventionists. One of the aims of this qualitative research was to consider the proposed mechanisms of action underpinning the intervention in practice. The mechanisms of action explain how the intervention components might work to produce an outcome [13,14] and how any theoretical model might be working within the intervention. Qualitative research can be undertaken during the development and feasibility phases of developing complex interventions [14,15] to help researchers understand how a complex intervention with multiple and interacting components might work [16]. Qualitative research within process evaluations can help researchers understand mechanisms of action by looking for unanticipated and complex causal pathways through the intervention or unanticipated consequences of the intervention [17,18], which is useful in refining the intervention before a full-scale evaluation.

The intervention is complex, with a number of interacting components [16]. A study by Arden et al (unpublished data, 2020) provides more details on the intervention and development processes. In summary, an eTrack nebulizer (PARI, Pharma GmbH) with electronic monitoring capabilities sends timestamped inhalation data to a 2net Hub (Qualcomm). This enables real-time monitoring of adherence via securely stored data on CFHealthHub servers. Adherence data are displayed on the CFHealthHub website and are visible to patients and health professionals delivering the intervention. The intervention also comprises a manualized behavior change component that is designed to help patients increase their adherence. This is underpinned by a theoretical model of behavior change (Figure 1) based on the idea that reflective motivation, in which people rationally weigh up the perceived necessity of a behavior against the perceived concerns they have about treatment, underpins adherence behavior. Therefore, some participants would need to address motivation before other strategies, such as action planning, could be successfully used. The model of behavior change is part of a larger logic model [12], informed by qualitative interviews with adults with CF about their adherence behavior [19,20] mapped to the Theoretical Domains Framework [21] and behavior change wheel [22] and developed using a person-based approach [23]. The behavior change intervention is delivered by trained health professionals (interventionists) in a series of face-to-face and telephone meetings with patients. Interventionists help patients to increase their adherence to nebulizer treatments by choosing components of the behavior change intervention to address an individual’s unique issues identified through baseline questionnaires and discussions during intervention visits and placed into a toolkit area of the CFHealthHub website. The interventionists in the pilot RCT were clinicians employed to deliver the intervention. The number and location of intervention visits are tailored to patients’ needs.

Figure 1 [24] shows how the model splits the behavior change components into components for all participants (including those with low motivation) and for those with adequate motivation (#1 to #7) and how those components link to mechanisms of action (#8 to #13) that lead to increased adherence. Table 1 describes the different components and the proposed mechanisms of action in more detail and how they are delivered within the intervention. The numbers in brackets in the table are linked to the numbers in Figure 1.

Whilst considering the proposed mechanisms of action in Table 1 in a qualitative study, we identified 3 mechanisms of action (relationships, visibility, and fit) that Vassilev et al [25] found in a realist review of effective telehealth interventions that explain how telehealth supports self-management of long-term conditions. First, effective telehealth interventions should consider how relationships with health professionals and peers are enabled or inhibited. Second, the intervention should enable the visibility of symptoms to the self and others and provide feedback to increase knowledge, motivation, and empower patients, thus enabling reinforcement of behavior change. Finally, successful interventions should fit, or at least not disrupt, the everyday lives and routines of patients, including their existing skills and knowledge. The study by Vassilev et al [25] is transferable to this intervention because there are key components of telehealth, namely, the intervention has web-based monitoring and web-based behavior change materials, the intervention is concerned with self-management (adherence to nebulizer treatments), and the condition being studied, CF, is a long-term condition. Although relationships, fit, and visibility are implicit in the stated intervention components, the ways in which they might work in the intervention as potential mechanisms of action were not fully articulated; thus, their importance to the potential effectiveness of the intervention needed further exploration. In this study, we explore the mechanisms of action operating in the intervention by confirming how the expected mechanisms of action identified in Table 1 worked in the intervention and by showing how important the mechanisms of action (relationships, visibility, and fit) identified in the review by Vassilev et al [25] of telehealth for self-management support were to facilitate the potential effectiveness of this intervention.

We undertook a feasibility study in preparation for a full-sized RCT. The study consisted of a mixed methods process evaluation undertaken concurrently with a pilot RCT in 2 UK CF centers. In total, 33 adults with CF received the intervention for 5 months [12]. The process evaluation addressed the feasibility and acceptability of the intervention and the RCT and the mechanisms of action of the intervention. The qualitative component of the process evaluation allowed exploration of the mechanisms of action at play, and this is the focus of this paper. The study received ethical approval from the London Brent Research Ethics Committee (16/LO/0356).

We conducted 25 semistructured face-to-face interviews, with patients in the intervention arm of the RCT (n=14), interventionists (n=3 at 2 time points), and members of the wider MDT offering care to patients with CF (n=5). We interviewed more patients in RCT site 1 (n=8) than in RCT site 2 (n=6), because fewer patients agreed to be interviewed at site 2. Sociodemographic data for patient participants are shown in Table 2.

We interviewed all 3 interventionists. They had different backgrounds: 2 had worked as physiotherapists within the multidisciplinary health care team who provided care to patients with CF in that center and one was a psychologist who had not previously worked in a CF setting. The MDT members were from a mix of disciplines, including respiratory consultants and physiotherapists.

During consent for the pilot RCT, 28 of the 33 intervention patients consented to be approached for an interview. We invited participants by letter or email and followed up by telephone or email. Of the 28 available patients, 16 consented to interview, 2 declined, 1 withdrew from the full study, 1 died during the study, we were unable to contact 7 patients, and we did not approach 1 patient because of time constraints. Of the 16 consented, 14 were interviewed and 2 patients canceled their interviews. The 3 interventionists consented to be interviewed twice during the study, toward the beginning and the end of the pilot RCT. MDT participants were recruited through the interventionists at each site. We obtained informed written consent from all participants before the interview.

Interviews were conducted by SD, AS, and SK. Patients were interviewed face-to-face in their homes or the hospital depending on preference, whereas interventionists and MDT interviews were conducted at the hospital, except one that was by telephone. We used topic guides as the basis for the interviews, with questions about the acceptability of different aspects of the intervention and RCT, and explored which aspects of the intervention participants perceived to be important in increasing adherence to nebulizer treatments (mechanisms of action). Interviews were also open enough to allow patients to describe their approaches to adherence and any challenges they faced. The interviews lasted between 11 and 102 min, averaging 56 min. Interviews were digitally audio-recorded and transcribed verbatim. We removed identifying data before analysis and gave participants an identification number to maintain anonymity.

SD and AS conducted the initial coding. We initially derived a coding framework [28] based on the Theoretical Domains Framework categories related to adherence [19] and the key aspects of process evaluation (context, mechanisms, and implementation) [17]. After initial familiarization with the data, SD added the telehealth mechanisms of action by Vassilev to the coding framework, because they were evident in the data [25]. SD and AS read the transcripts and coded them against the framework using NVivo software, allowing for other themes to be identified in the data set that did not fit the framework. After the initial coding, SD worked with MA to discuss the relationship between the telehealth mechanisms by Vassilev (relationships, visibility, and fit) and the expected mechanisms of action based on behavior change theory. Trustworthiness was addressed through: researchers writing reflections after each interview, discussion of the interviews between SD, AS, and SK during data collection, discussion of the findings between SD and AC to develop the analysis, and presentations to the wider research team. We present quotes from patients in the intervention (P), interventionists (INT), and members of the MDT to provide examples of the points we are making.

A trained interventionist delivered the intervention through a series of tailored meetings where patients and interventionists discussed all aspects of adherence behavior. The way the interventionists communicated was intended to encourage open and honest communication about adherence, from which a realistic understanding of the barriers that patients faced in their daily lives could emerge. The importance of the style of communication was addressed in the intervention manual and during the interventionists’ training program. In the pilot RCT, patients and interventionists emphasized the importance of these interactions between interventionists and patients in helping them to improve their adherence. In particular, they valued where the interactions occurred and how interventionists communicated with patients.

Meetings in a patient’s home were face-to-face, lasting up to an hour. These meetings were highly valued by some patients because patients perceived that the time given and effort made to travel to a patient’s home by the interventionist were indications of their interest in the patient’s adherence:

The fact that visits were home-based was also important to some patients because this venue provided a safer space in which patients had time to talk in-depth about adherence. Interventionists also appreciated being able to find out about the lives of patients and some of the barriers they faced by seeing their living conditions. For some patients, this was an important first step in reflecting on their adherence behavior and trying to understand and identify potential barriers. It also demonstrated to patients the importance of focusing on their adherence:

Face-to-face meetings also gave the opportunity for informal talk, which helped to establish a rapport between patients and interventionists. One interventionist felt that the script-like nature of the intervention could prevent relationship building, so they tried to make the intervention visits flow by including general chit-chat and letting patients make conversation at their own pace. This contributed to patients feeling like the interventionists displayed a caring attitude toward them:

This caring attitude was an important aspect in building a relationship that had the potential to help the patients increase their adherence by allowing them to be open about the barriers they faced to adherence.

Visibility was a key aspect of the intervention. Self-monitoring of real-time adherence data in the form of graphs and tables against personalized goals, in combination with open discussions about adherence behavior (see Relationships: Building a Relationship With the Interventionist), was intended to raise the visibility of adherence for patients and clinicians by aiding memory about specific instances of adherence or nonadherence. Discussions could then focus on identifying barriers to adherence, which could be addressed to improve self-efficacy. On the other hand, although monitoring of data by the multidisciplinary health care team was part of the intervention, it was not planned as a central component of the intervention and was not linked to a mechanism of action to increase adherence directly. In the pilot RCT, it was apparent that visibility to the self was an important motivator to improve adherence while visibility to others, that is, people they felt accountable to, could also be a motivator.

In the pilot RCT, we found that visibility to the self-operated as expected through self-monitoring (Table 3). Visibility to others operated when the intervention allowed the interventionist, and with patient consent, their multidisciplinary health care team, to see adherence data. Although most patients we interviewed chose to share their data with their health care team, some patients did not always like being monitored by others. Indeed, some members of the multidisciplinary health care team worried that the intervention could be perceived as unwelcome, described as Big-Brotherish by one interviewee (MDT1, site 2), involving the health care team checking up on patients or telling them off.

Patients who chose to share their data described how others monitoring adherence could make them accountable to someone else in their health care team. They also described how others’ monitoring motivated them to adhere through a desire to look good in front of others:

Offering praise to patients to increase their adherence could further motivate patients. For example, although this was not specified as part of the intervention, one of the interventionists would text patients with praise or encouragement to keep going if graphs showed improvements:

The intervention was developed to be tailored to individual patients so that the education about treatments, setting goals and action plans, and problem solving fitted patients’ specific needs. Particular content identified by the interventionists could be added to the toolkit area of the website, offering a tailored space for each patient. In addition, the intervention was tailored to patient circumstances by offering options for interventionist visits in the patient’s home, hospital, or by telephone.

In the pilot RCT, patients valued this tailoring and believed that it facilitated adherence improvement because not every patient had the same experience or background, so it was important that the intervention allowed patients to focus on the aspects that were most useful to them:

Although patients appreciated the ability to tailor the intervention, it was important to them that this tailoring was patient-led, that is, tailored to the patient’s needs rather than the interventionist suggesting or even dictating what targets and goals should be set by the patient:

One patient with high baseline adherence appreciated how the interventionist recognized that they did not need to set a goal to improve adherence, that is, they did not believe they needed some aspects of the intervention:

However, there were problems around fit with patients’ perceived needs and lives, with some of the patients seeing the intervention as a better fit for others rather than themselves. Although most patients in the sample believed the intervention was useful, some patients with baseline high adherence did not see the need for setting action plans, but believed that those struggling with adherence would benefit from these. Some patients also perceived that education, problem solving, and video components were more useful for other people, such as younger or newly diagnosed patients with less knowledge:

Importantly, some patients believed that the intervention did not fit their lifestyles because they were busy and could not access the intervention away from their desktop. For example, although the toolkit area was appreciated by patients, interventionists noticed that they were unlikely to look at the materials outside the meetings. Patients described not having time to log into the website and preferring an app to access the website flexibly when away from home. There was also an issue around fit with priorities in life. For one patient, adherence was low down their list of priorities compared with work, although there were signs that the intervention had started to help them challenge that belief:

There was a view among health professionals that the intervention fitted the needs and priorities of patients with moderate baseline adherence rates because high adherers were adhering without the intervention, and low adherers, who perhaps had the most need, were difficult to engage:

In this study, we described how components and mechanisms of action for behavior change planned within a new intervention potentially helped patients to increase adherence to nebulizer treatments. We also described how mechanisms of action for effective telehealth interventions for self-management support (relationships, visibility, and fit) [25] were operationalized and valued by patients using the intervention. Patients described how building a relationship with the interventionist through face-to-face visits in the home with someone who cared about them and their progress helped them to consider ways of increasing adherence to medication. Rather than seeing the visibility of adherence data to clinicians as problematic, patients found this motivating because it made them accountable to someone else, particularly if they received praise about progress made. Although the intervention was tailored to individuals, there were challenges in how the intervention fitted into some patients’ busy lives, particularly when delivered through a desktop computer. The intervention was perceived as a better fit for other patients, for example, those who were younger or less knowledgeable. Interventionists identified a middle group of patients with moderate baseline adherence who might benefit more from the intervention.

Sitting down face-to-face with an interventionist provided participants with someone who they felt cared about them, providing motivation to increase their adherence. This echoes previous research that has shown how the human aspect of a telehealth intervention helped to build rapport and trust that could help a patient to change their behavior [29]. Although other research has found that self-management may be facilitated by clinic visits [30], in this study, interventionists and patients preferred home visits because they showed that the interventionist was invested in their care, put them at ease, and allowed them to focus on adherence.

The intervention worked through making adherence visible to patients, thereby empowering them to change behaviors. In terms of visibility to others, our study supported research that health care practitioners can be more concerned about negative aspects of surveillance than patients [25], although there was some evidence that those wishing to keep adherence private opted out or withdrew from the pilot RCT. Mohr et al [31] coined the term supportive accountability to describe how a trustworthy individual with expertise can support engagement with digital interventions [32]. In our study, this supportive accountability went beyond facilitating engagement and appeared to facilitate outcomes directly. Patients in our study appreciated praise from clinicians when they improved their adherence. Praise is a part of rewards that can facilitate adherence [19,30] by building motivation and self-efficacy [33] and by providing a supportive environment for behavior change [34]. Research has suggested that rewards may work best in combination with identifying barriers and problem solving [35], which both occurred in this intervention.

Vassilev et al [25] suggested that a fit to everyday needs, skills, and daily life was important for intervention success. Research suggests that the fit may help to increase autonomy and thus motivation to change behavior [32]. Our finding that the intervention sometimes did not fit into people’s busy lives is supported by research exploring barriers to adherence in CF where people often cite being too busy to adhere to treatments [19,36-45] and social and work demands creating competing priorities [19,36,46]. This highlights the importance of offering the intervention on an app to enable patients to access adherence data away from their desktop. Some patients also perceived that the educational aspects of the intervention fit better with the needs of younger or newly diagnosed patients who had less knowledge of nebulizer treatments. Research with adolescents and young adults has suggested that they may face similar barriers to adherence as adults [36,38-40,42]. Although the educational aspects of the intervention, such as the videos of how treatments work, may have the potential for use with that group or for those who receive a later diagnosis, we would argue that education alone is unlikely to be sufficient to increase adherence.

We also found perceptions that the intervention fits some groups better than others based on their adherence level with a group of moderate adherers seeming to benefit and engage more than low or high adherers. This could support research into grouping patients by adherence level [11,27], which may have implications for health outcomes [11].

This study had some limitations. The sample of patient participants at one of the sites was small because fewer patients agreed to be approached for interview when they were recruited for the pilot RCT. Despite this, we have provided an understanding of patients’ experiences of a new intervention and how the mechanisms of action operated in practice.

We modified the intervention based on the results of the wider process evaluation, including the insights raised here [12]. Key implications for the intervention were that the interventionist was a very important aspect of the intervention; there needed to be a feedback mechanism to reward patients for meeting their targets in order to maintain motivation, and that the intervention needed to fit into busy lives, highlighting the need for an app that was produced for use in a future full-scale RCT.

This behavior change intervention with telehealth components had the expected behavior change mechanisms of action and mechanisms of action associated with effective telehealth interventions for self-management support. The intervention was modified to strengthen mechanisms of action based on these findings, for example, delivery through an app accessed via mobile phones, which is ready for testing in an RCT in 19 UK CF centers.