The study focused on gathering information from patients and professionals about the use of PP to develop an app containing PPIs for patients with BD. A qualitative design was used to identify and characterize opinions and needs. The target group comprised patients with BD and professionals (end users). To enable a broad perspective on the potential benefits of web-based PPIs for patients with BD, focus group discussions (FGDs), paper prototyping (PPT), and pilot test (PT) were used. The method was modeled on the principles of the holistic development approach described in the Center for eHealth and Disease Management (CeHRes) road map, for example, the participatory process and continuous evaluation cycles [70]. Our study covered the first 3 steps of the model: contextual inquiry, VS, and design. The CeHRes aims to facilitate continuing process of evaluation and participation of all stakeholders.

Ethics approval was obtained from the University of Twente (18067) and the scientific board of the Dimence Mental Health Institute, where the study was conducted. All participants signed an informed consent form, in which they also agreed to be audio recorded during the focus groups (FGs).

Participants were patients with BD I or II in a euthymic episode and professionals treating patients with BD who were in possession of a smartphone and were willing to travel to attend the FGs. They participated in all 3 steps of the development process. When recruiting our sample, we aimed for maximum variation among the participants of the FG. Therefore, we used several recruitment strategies. Patients were recruited from an outpatient clinic and the national advocacy group (Plusminus) to receive input from different regions of the Netherlands. In using this method of recruiting, we tried to avoid selection bias. All the patients received treatment as usual for BD. A call for participation was posted in the advocacy group’s magazine. The researchers who participated in the FGs (BG and SK) were not involved in the participants’ treatments. Professionals in the FG were involved in various disciplines (psychiatrists, psychiatric nurses, and psychologists) and recruited from the Dutch Foundation for Bipolar Disorders (Kenniscentrum Bipolaire Stoornissen), a chapter of the International Society for Bipolar Disorders.

Participants contributed to all study phases (FG, PPT, and PT); however, not all participants attended every FG. None of the participants in the FG had professional relationships, which means that none of the patients were treated by the professionals or otherwise involved with each other.

ATLAS.ti7 was used for the analysis of the data from the FG. The Colaizzi method, as described by Shosha [72], was used to process the data. The FG recordings were transcribed verbatim for the analysis in 3 phases: open coding, axial coding, and selective coding. Researcher triangulation was used to increase the objectivity of data analysis. The app’s quantitative data (rating of the exercises) were collected, and average scores for each exercise and all the 7 exercises were calculated using SSPS. In addition, we also calculated the average scores of the differences between professionals and patients. The qualitative data (open answers in the evaluation module) were collected and analyzed through inductive coding.

For the study, 17 participants (n=11, 65% patients and n=6, 35% professionals) were recruited. In total, 24% (4/17) of the participants withdrew (3/4, 75% patients and 1/4, 25% professional) before the start of the study owing to personal reasons or because they were not available at the time of the FGs. For the last phase (ie, PT of the app), 6 participants were added to broaden the input with opinions of participants who did not participate in the FGs. Demographics are shown in Table 2. Table 3 shows the participation in the FG, RPT, and PT.

The results of all FGs are presented based on the different stages of the CeHRes road map. The results of the contextual inquiry and VS are summarized to provide a good overview of the FG results.

In the FG, first, we discussed the participants’ level of experience with PP. Second, we created an inventory of expected advantages and disadvantages when PPI is applied for BD. The results are summarized in Table 4.

Among the FG members, there was little experience with PP, as shown in Table 4. The participants did not have experiences with specific evidence-based PPIs. The experiences can be found in adjoining therapeutic areas (eg, mindfulness) or more personal PP solutions (eg, recognizing positive moments or thinking about possibilities rather than about problems). The FG members mentioned possible advantages: focusing on small steps, making positive pictures or movies, monitoring positive feelings, and giving themselves a positive message. Unmet needs such as hope, acceptance, and increasing self-confidence appear to be the most promising ones that PPIs may address in BD. The participants also mentioned various potential disadvantages of PPIs in 2 categories: illness-related and personal factors. The expectation of disadvantages regarding their illness is seen in both mania and depressive episodes. The participants did not expect benefits from PPIs in severe depressive episodes, or even the inability to perform the exercises during severe depression, leading to disappointment rather than satisfaction. They also foresaw further mood dysregulation toward manic stages when they are already hypomanic owing to exercises that stimulate positive emotions, feelings, or happiness. On a personal level, the participants fear forced positive statements that are not consistent with their self-esteem and the risk that PPIs can push the goal setting level very far (perfectionism). For both mentioned domains, they fear a possible counterproductive outcome when PPIs are applied.

The aim of the FGs was to address opinions about the use and requirements of an app. First, the participants were asked about their opinions on the different categories of PPIs (Table 5). The FG participants were unanimous in the appreciation of the exercises in positive emotions and resilience categories and found positive relations and strengths to be applicable for patients with BD. In the categories of self-compassion, hope, and optimism, the FG participants were less convinced; there was fear that they could not fulfill high expectations of themselves or the app demands. Second, a recurring topic in the discussion was personalization; individuals can have different preferences for exercises. Therefore, all types of exercises should be available within the app. In addition, the individual user can choose certain types of exercises. Third, some participants found it important that the choice was made in alignment with the professional caregiver. Fourth, we concretized the app’s design on 2 levels: the app’s use (VS) and its feel and look (design).

The most prominent subtopics in the use of the app were the following: when the app is used and under what conditions. Furthermore, the FG members made suggestions for more advanced use. These results are summarized in Table 6. The participants were unanimous that the best way to use OPPIs is during euthymic or mild depressive episodes. Participants were divided over the use of an OPPI in episodes of mania or hypomania. The FG members expected that use in the early stages of hypomania can provoke positive feelings and lead to a more severe manic state. However, when users are in a full manic state, they did not expect any temptation to use the app in a full manic episode because, when manic, they are quickly distracted, and the exercises require time and tranquility.

The participants also indicated that the app should be adjustable; however, most FG members think that they want to use it daily, during periods when it is beneficial to them. The duration of the practices should be between 5 and 10 minutes, so that it fits into daily routines. The participants also mentioned that allowing sufficient time and having a quiet place is important for successfully using the app. In addition, the FG members suggested that, after performing a complete set of practices, the user can choose which practices are suitable and adjust the app to those preferences. In the discussion about the potential advanced possibilities, the FG members found it helpful to connect PPIs with the Life Chart Method (LCM) [73]. A PPI should occur when an advanced set value is reached while monitoring symptoms such as mild depression (within the LCM). Incorporating PPIs in the early relapse prevention plan was also suggested, as users can apply this intervention to handle the (starting) mood episode. In the ideal world, the FG members want to be automatically directed to the PPI when they reach pre-established levels in the LCM.

During FG 2 and 3, participants were asked about their views on the requirements for the design of the app. The remarks obtained can be separated into the following subcategories: personalization, look and feel, text, vision and sound, and preferred options. Table 7 summarizes the results.

The participants considered personalization to be a vital aspect. The FG members wanted to have a wide range of adjustable options as long it does not affect the app’s clarity. The main topics in personalization were the frequency of use, choice between reading or listening, notifications, writing space in the exercises, and possibility to select the exercises. In addition, the proper use of wording was considered necessary by the participants. The text in the app should be concrete, clear, and short. Within the suggested exercises, the choice of words was sometimes perceived as compelling. Examples to illustrate the exercises were seen as helpful and supportive. The FG members had a preference for appealing messages in both exercises and notifications. They preferred a design that is quiet and attractive. The use of pictures can have a calming effect. We also discussed the option of obtaining feedback after completing an exercise. Approximately all participants (7/8, 88%) liked some type of feedback. They also expressed the need for external motivation to continue practicing (apart from the notifications). Significant others can be part of this motivational aspect. Valuing the exercises was seen as a good instrument to decide which exercises were preferable.

As part of the VS and design, we conducted an RPT test. Owing to technical problems in registration, 25% (2/8) of the participants did not succeed. The feedback obtained from the participants was divided into 4 categories: experienced effects, facilitating factors, impediments, and suggestions. Regarding experienced effects, 75% (6/8) of the participants found exercises to be beneficial and experienced more positive emotions than before. The experienced positive emotions did not qualify as ‘threatening’ in terms of risk for a (hypo)manic episode. One participant did not experience any difference, and another participant found it difficult to perform the exercise owing to personal circumstances. Regarding facilitating factors, the participants appreciated the layout with different pictures, videos, and music and the calm design. In addition, participants emphasized the possibility of adding the option to personalize the look and feel of the app. In total, 38% (3/8) of the participants experienced impediments while testing the app. Overall, 25% (2/8) of the participants mentioned that they were very severely depressed to perform the exercise.

On the basis of the consensus reached in the FG and results of the RPT, we developed the Wellbeing Bipolar Disorder app. The app contains 7 exercises in the 4 domains of PP preferred by FG members. The exercises based on the previous work of Bohlmeijer and Hulsbergen [74] are shown in Table 8. The app differs from other apps that provide PPIs, primarily owing to the selection of PPIs. Some exercises were withdrawn because of the fear of compassion or had to be altered. Other exercises were withdrawn because they may provoke symptoms of mania. Moreover, the app had special design features, such as calm design, proper use of wording, and so on, which, according to the FG members, can lead to high compliance rates. Finally, the app can be integrated into the LCM.

We chose a 1-week period in which the participants were provided 1 exercise daily, with mood monitoring before and after exercise and a valuation of the exercise after completing it. The app was personalized with the following choices: when to use the app, notifications, types of notifications, written or spoken video explanation, and guidance by a professional or expert by experience. The app’s design was calm. All exercises had 1 picture throughout the practice (Figure 1).

The participants were asked if they had succeeded in the exercise; if they did not, they could start again. We wrote 2 different scripts (professional or expert by experience), and the videos were set in the background picture belonging to the exercise.

After the development process, we tested our app to evaluate whether the results of the previous steps of the development process have been implemented to the participants’ satisfaction. The results of the PT are shown in Tables 9 and 10.

Of the 133 exercises (7×19 participants), 87 (65.4%) were completed. Patients completed more (101/133, 75.9%) exercises than professionals (72/133, 54.1%). In total, 11% (2/19) of the participants chose the video explanation. Of the 19 participants, 8 (42%) preferred the expert by experience and 8 (42%) preferred the guidance by a professional. Overall, the average rating of all exercises in total was 7.35 (scale 0-10, SD 0.525), and the median was 7.5, with a slightly high rating among professionals (mean 7.7 vs 6.9; median 7.5 vs 7.25).

The evaluation of the individual exercises was between 7.5 (exercise: be strong and be stronger) and 6.6 (exercise; listing to good news). Notably, a participant rated all exercises relatively low (average 3.3), and owing to the small sample size, this influenced the total outcome. The other individual ratings were between 8.7 and 5.6.

The app’s overall valuation was high; 91% (15/16) of the participants were positive about the app and wanted to use it for an extended period. However, according to 55% (9/16) of the participants, the frequency of the exercises seemed very high for an extended period. Besides positive comments, made in the app’s evaluation, about the effect of the exercises, there were remarks for improvement of the app. Some comments referred to the intensity of the exercises; a new exercise every day is not doable for all participants. Another advised option is to read the exercises in the morning to accomplish it during the day.

This study investigated the opinions of care professionals and patients with BD regarding OPPI for BD.

The first aim was to rate the extent to which participants had experience with PPIs and expected them to be supportive in addressing unmet needs. We found that the participants did not have experience with specific evidenced PPIs; however, they mentioned experiences in adjoining therapeutic areas (eg, mindfulness) or more personal PP solutions. The FG members mentioned possible advantages: focusing on small steps, making positive pictures or movies, monitoring positive feelings, and giving yourself a positive message. These topics can be related to various unmet needs in treating BD, specifically, offering hope, increasing self-esteem, expressing feelings, promoting acceptance, and preventing social isolation. Regarding expectations about PPIs, participants expect that PPIs can accomplish some unmet needs in BD.

The second aim was to establish the preferences and requirements of the app with PPIs.

First, we determined which PP exercises patients and professionals prefer. Positive emotions, resilience, positive relations, and strengths were highly valued among the 6 categories of PP exercises. Some exercises were withdrawn due to the fear of compassion or had to be altered. Other exercises were withdrawn because they may provoke symptoms of mania. Moreover, the app had special design features, such as calm design, proper use of wording, and so on, which, according to the FG members, can lead to high compliance rates. Finally, the app can be integrated into the LCM. Therefore, we developed an app that differs from other apps that provide PPIs, primarily owing to the selection of PPIs.

The third aim of our study was to evaluate the use and satisfaction of the app. An interesting finding is that approximately all users (15/16, 91%) found it to be beneficial to perform the exercises and wanted to do it regularly. However, the frequency of the exercises seems to be very high. The valuation of the exercises was promising (7.35 on a scale of 1-10; median 7.5). Despite the small number of participants, we seemed to have found the proper exercises for our target group. Before releasing the app for clinical practice, further studies with adequate measurements (quantitative and qualitative) are necessary.

Unmet needs can include topics such as support with loneliness, grief counseling, acceptation, social isolation, coping with others in social functioning [14,15,17], hope, expressing feelings, and increasing self-confidence in psychological functioning [12,14-17]. The positive expectations underscore the potential of PP for personal recovery and support the value of integrating PPIs into mental health care [57,58]. PPIs may help patients with resources that increase their ability to adapt and support them in personal recovery [24]. Our results are largely consistent with previous findings [19,75,76]. Mental well-being is recognized as an essential resource for the recovery from mental illness and in preventing relapse. Therefore, it is recommended to include mental well-being interventions (such as PPI) in the treatment [75,77].

Participants also raised some critical concerns about applying PPIs for BD. The FG members expressed concerns about fast and more severe changes in mood and energy when a used PPIs during manic or hypomanic episodes. Although “joy and amusement” are associated with increased manic severity, compassion—one of the key elements of PP—tends to decrease the symptoms of mania [78]. Positive emotions are not often mentioned as triggers that can provoke manic episodes. Lack of sleep is the most predominant factor in triggering manic episodes [79]. However, amorousness is the predominant factor among young adults, followed by stressful life events [80]. Periods of strong personal growth are also factors that induce manic symptoms (such as an extremely motivational workshop) [80]. Although this seems to be linked to PPIs, it is not satisfactory to conclude that PPIs can induce manic episodes. In a study among a large population of patients with BD (n=149), by applying a group PPI, the researchers found no increase in manic symptoms, thus confirming our findings [55]. In contrast, damping emotions lead to more severe depressive symptoms [81]. Therefore, it is recommended to inform potential participants properly before applying PPIs for BD. Interestingly, the PT results did not show dysregulation among the participants.

An unexpected finding was that the categories of self-compassion, optimism, and hope had low ratings. On the basis of the unmet needs, as described previously, we did not expect this outcome. This result may be explained by the fact that the FG members found the themes and exercises to be very ambitious. For example, the exercise, “the best possible self,” seems not to fit with the participants’ level of self-esteem. They had high expectations that they could not satisfy.

The “fear of compassion” can explain the fact that the FG members valued compassion relatively low. The fear of compassion is closely linked to self-criticism and depression [81]. Women with BD seem to be more self-critical than controls [82]. Nitzburg et al [83] suggest that the negative experiences in the course of BD can worsen the level of self-criticism and argue that in an early stage of the illness, interventions should also target decreasing self-criticism. However, owing to the small sample size, we cannot rule out that our findings are coincidental. Nevertheless, it is necessary to develop interventions targeting compassion and hope in BD to pay attention to the fear of compassion and transform exercises in a feasible manner for patients with BD.

We determined when to use the app. Some of the FG members expect the exercises to be a daily routine, for example, in addition to mood monitoring. Integration with a (digital) mood monitoring application and relapse prevention plan can provide a comprehensive tool in which the PPIs assist in preventing severe mood episodes [84]. The FG members foresaw the use mainly during euthymic or mild depressive episodes. The members of the FGs were unanimous in that they did not expect any beneficial effects from PPI in severe depressive episodes. In contrast, they expected the opposite effect: worsening of depressive feelings if the goals of the exercises are not achieved. Previous meta-analyses have demonstrated that PPIs seem to be more effective in populations that are nondepressed or mildly depressed [28,50] than in populations with major mood disorders [50]. This supported the FG members’ suggestion to apply PPI only in mild depressive episodes and euthymic episodes. Carr et al [85] published a meta-analysis showing that in patients with depression, PPIs had a small to moderate effect on depression in terms of symptom reduction. However, in that paper, the severity of the depression was unclear. The considerations mentioned previously can underline our finding that personalization is one of the main topics in designing a PPI app for BD. Lack of personalization is, among with depressive symptoms, an important barrier to adapting digital mental health interventions [86]. Recognizing and resolving these barriers in the development process can contribute to the adaptation of new interventions [86].

Finally, we reflected on the method. We recognized that cocreation and user involvement are important in developing an app to be accepted by end users [87]. Specific findings, such as the preferences for the exercise categories or the possible risks in applying PPIs, will not be revealed, as we developed the app without user involvement. The participation of end users is the main element in user-centered design, defined by Preece et al [88] as “an approach, which views knowledge about users and their involvement in the design process as a central concern.” The development process conformed to the CeHRes principles [70]. This method allowed us to systematically involve users (patients and professionals) in the study’s development process. This enabled us to modify the app’s design through all stages of the development process and guaranteed maximum involvement of all stakeholders. Previous studies using the same principle support our findings [88]. The development of new interventions benefits from user involvement in all stages to meet the target group’s needs [89].

Our study has some limitations. When recruiting professionals, there is a risk of selection bias owing to the voluntary recruitment from a highly specialized professional pool (the Dutch Foundation for Bipolar Disorders). It is conceivable that the professionals interested in web-based monitoring are stepping forward to participate in this study. However, this method was chosen to obtain professional input from different parts of the Netherlands to avoid inputs from only one region. The same applies to the consumers; they were partly recruited from an outpatient clinic and partly from the Dutch advocacy network, “Plusminus.” Although this may raise the question of whether the participants represent the target group, we assume that, by combining advocacy members and patients treated in an outpatient clinic, our study is sufficiently representative.

The use of FGs has some limitations, such as the possibility of “group effect,” in which patients tend to adapt to the group leading opinions. Therefore, it is difficult to separate a personal opinion from a group opinion [90]. We tried to avoid this bias by collecting individual data (post-it memos) before the discussion in the group. Furthermore, it is sometimes difficult to generalize the outcome of FGs [90]. We tried to avoid this bias as much as possible through nationwide recruitment.

Finally, this study is mainly a qualitative study with a relatively small number of participants; therefore, it remains uncertain whether the results are sufficiently generalizable.

Despite these limitations, we believe that we shed light on consumers’ and professionals’ thoughts and considerations about using PP apps for BD.

Despite recognizing the possible benefits of PPI in BD and that they may address unmet needs in BD, very little is known about the effect of applying PPI in the treatment of BD.

In this study, we realized the shared assumptions about the application of PPIs for BD. The consensus on the different topics regarding the use of PPI shows that both patients and professionals underline the beneficial possibility of applying PPIs for BD. The use during subsyndromal and mild depressive episodes seems to be the most fruitful period for patients with BD. We did not establish the risk of provoking mania or hypomania by performing PPIs, but we could not draw firm conclusions because of the small sample size.

With the systematic use of user involvement (patients and professionals) in all steps of the development process, we were able to create an app that can potentially fulfill some of the current unmet needs in the treatment of BS.

The use of PPI for BD is intriguing and can be usefully explored in further studies. We emphasize that more evaluation studies (quantitative and qualitative) that are focused on the effect of PPIs in the treatment of BD should be conducted. In addition, to establish the working mechanisms in BD, explorative, qualitative, designed studies are required to reveal whether PPIs can address unmet needs in BD.