User involvement in the development of health care services is important for the viability, usability, and effectiveness of services [1-3]. However, even when the users are involved, the development process may be unsuccessful, and the service may not be accepted and used. Previous research has reported diverse results regarding both benefits and drawbacks of user involvement [1,2]. One explanation for the drawbacks may be that little is known about the various aspects of collaboration associated with the development of a successful service [4,5].

Co-design implies active involvement of lead users (ie, potential users of the future service) to incorporate their experiences and knowledge into the new service [6]. “User-centered design” [7] and “participatory design” [1] are umbrella terms for design strategies. Experience-based design (EBD) [6] is a subform of participatory design that involves including lead users as co-designers. EBD uses principles from design science including architecture, product, and computer design to make the service or product safe, effective, and enjoyable for the user. Because the users’ experiences are essential to providing optimal care, they have been used to improve health care services [8]. EBD is also complementary with personalized care, which focuses on the individuals’ experiences, preferences, and goals in the provision of optimal care [9,10].

Co-design also denotes a collaboration between the stakeholders and system developers. Collaboration can be defined as the interaction between the stakeholders and system developers with the aim of achieving a shared goal [11]. Successful collaborations may be a key factor in the outcome of a project [12,13]. However, previous research related to co-design has primarily described the process on a macro level, for example, the benefits and drawbacks regarding money, time, and how user participation informs the new service [2,14]. Empirical studies describing the collaboration between co-design participants are scarce [2,4,5]. We were able to identify only 1 previous study that used video recordings to describe the collaboration during co-design meetings [15]. Therefore, research describing the collaboration during co-design is needed to extend our knowledge on effective collaborations to move the process forward.

This study will provide a description of the challenges observed during co-design meetings. During the first step of the ongoing project, lead users presented ideas on core features that are to be included in the future service [16]. During the second step of the co-design, lead users, a clinical physiotherapist, researchers with knowledge in physical activity (PA) in rheumatoid arthritis (RA) and behavioral learning theory, an eHealth strategist, and an officer from the patient organization collaborated to provide data for the system requirements and specification of the future service [17]. The aim of this study was to explore the challenges in advancing the co-design process during the requirements specification phase of a mobile Internet service for the self-management of PA in RA.

To explore the challenges in advancing the co-design process, an inductive, qualitative, participatory-action research design was applied [18]. The co-design process was performed during 4 workshops in February and March, 2013, at Uppsala University, Sweden.

Data collection included video recordings [19] and naturalistic observations [20]. The purpose was to capture situations and events in which the co-design participants discussed issues deemed important for advancing the co-design process. This study was approved by the regional ethical review board in Stockholm (D nr 2010/1101-31/5).

The co-design group (n=10) was formed to create a feasible workgroup and to capture different perspectives (ie, experiential and theoretical knowledge). Potential co-design participants were identified through our research and clinical networks and were invited by email by the first author. The inclusion criteria were adequate Swedish communication skills and access to the Internet with confidence in using the Internet. Furthermore, the participants were chosen to include different perspectives, including experiential knowledge in living with RA, clinical experience in supporting individuals with RA to be physically active, theoretical knowledge on behavioral learning theory, evidence for PA in RA, and/or service design.

Potential co-design participants who provided preliminary consent were informed about the study by the first author. Written information and a questionnaire on background characteristics, expertise, PA behavior, and Internet habits were provided by mail or email. Participants provided their final consent for participation by attending the first workshop. All but 1 of the participants agreed to attend all 4 workshops.

The participants included 5 lead users, including a patient research partner, 2 researchers with knowledge in behavioral learning theories and PA in RA, 1 clinical physiotherapist, 1 officer from the Swedish Rheumatism Association, and 1 eHealth strategist. Three of the participants were men, and the median age was 55 years (age range 34-73 years). All but one of the participants possessed a university degree. The lead users were chosen to reflect diversity regarding age, sex, years since diagnosis, and PA habits. A few of the participants reported experiential and/or theoretical knowledge in more than 1 of the perspectives (ie, experiential knowledge in living with RA, knowledge of behavioral learning theories and evidence for PA in RA, clinical experience in supporting individuals with RA to adopt and maintain PA, and/or service design).

A pilot workshop was held before the start of the study to test the data-collection procedures; for example, technical solutions for the video recordings and the feasibility of an observation protocol used by the observers. This resulted in decisions on how to arrange the participants’ seating and where to place the microphone and camera. In addition, it was decided that a technician was needed to set up the camera before each workshop to integrate the video recordings with the interactive.

The positioning of the camera can have a significant impact on the captured data and was carefully considered [19]. The camera was placed on a tripod and was positioned to capture the collaboration between participants, their faces and nonverbal actions, and references to the mediational means used (ie, interactive boards, an online notice board (Trello), and plastic sheets with an outlined mobile phone) (Figure 1). The interactive board and plastic sheet facilitated visualization of the future mobile Internet service during the discussion. They were also used to collect data on the system requirements and specification of the future service. These data were analyzed and presented elsewhere [17].

To facilitate and collect data during the co-design workshops, 1 moderator and 3 researchers were present. The moderator, who had substantial experience with moderation, programming, and designing of digital devices, directed the workshops. The last author (PÅ, experienced in qualitative research and research on physiotherapy integrated with behavioral medicine) was responsible for alignment of the process. The camera operator (CM, experienced in qualitative research and video-based research) and 2 observers (ÅR, experienced in qualitative research and behavioral medicine; CK, experienced in qualitative research, naturalistic observations, and health informatics) collected data during the workshops. The observers only watched and took notes and did not participate in the workshops (Figure 2).

The 4 co-design workshops were performed at intervals of 1-4 weeks in university lecture rooms and lasted between 3½ and 5½ hours. The aim of the workshops was to provide basic data on the system requirements and specifications for the mobile Internet service. The first workshop started from the results of the first step of the co-design process: core features of a future Internet service as proposed by lead users participating in focus group interviews [16]. Thereafter, the first and last authors (ÅR and PÅ), the eHealth strategist, and the moderator planned the workshops iteratively; that is, each co-design workshop was built on the results and experiences from the previous workshop. Textbox 1 presents an overview of the major content and participants at each co-design workshop.

Data were collected by video recordings [19] and an observation protocol [20].

The data consisted of approximately 16 hours of video recordings along with the observation protocols. The video recordings enabled repeated viewings of the relevant sequences where the challenges occurred. It also gave the other researchers the opportunity to discuss, confirm, reject, or adjust the analysis presented.

The inductive video-based analysis [19] was performed (ÅR) in 8 major steps (Textbox 2). The transfer of video data into text data resulted in a more thorough understanding of the challenges. A description of the transcript symbols and abbreviations used for participants are provided in Tables 1 and 2, respectively [21]. In addition, repetition of speech was reduced, and all the participants are referred to as female and the moderator as male.

Researcher triangulation was used to ensure trustworthiness during the analytic process. The first author participated in regular meetings with the second author, who guided the first author in the video-based analysis. Discussions involved deciphering what occurred during the selected sequences, including the challenges and their consequences for the process. Furthermore, labeling the theme, subthemes, categories, and subcategories evolved. The categorization and illustration of issues were discussed with the last author 3 times during the analysis process, which resulted in a consensus of the categorization and further refinement of the aim.

The analysis resulted in 1 overarching theme and 2 subthemes. A total of 7 categories and 12 subcategories illustrated the challenges deemed important for advancing the co-design process toward the goal (Figure 3 and Table 3). The results will be outlined by descriptions of the content of the categories and subcategories. To further illustrate the challenges, excerpts of the transcripts will also be presented.

The participants struggled to merge their individual perspectives during all phases of the co-design process. The participants shared, argued, and considered their different viewpoints, and integrated and counterbalanced these differences to find mutual agreements and solutions. In addition, 2 essential areas of merging were identified: “Finding a common starting point” and “Deciding on design solutions” (Figure 3).

In this study, we learned that co-design is a complex and demanding process that faces several challenges. The challenges observed were related to reaching agreements and making decisions necessary to advance the co-design process. The merging of participants’ different perspectives was particularly challenging, yet it seemed to be the core activity in a successful co-design process.

The constant need to merge participants’ respective perspectives placed extensive demands on collaboration. Merging of perspectives can be described as the participants’ efforts to consider and value their different viewpoints and to integrate and counterbalance these into a mutual agreement to progress the co-design process toward the overall goal. The merging was built on individual perspectives, that is, experiential and theoretical knowledge and participants’ ability to share these with others. Merging also included finding solutions. Data showed that the co-design group exhibited 2 methods of negotiating and finding solutions: consensus solutions or solutions despite remaining disagreements. This description of the co-design group’s ability to merge is in accordance with what theoretically has been described as a “collectivity-of-practice” [22].

According to our results, the participants’ different points of view caused the process to be difficult and time consuming, which may lead to conflicts between lead users and system developers and negatively impact the process [23,24]. Relationship conflicts have a negative impact, whereas task conflicts improve group performance [25]. Our data indicated that the co-design group primarily experienced task conflicts. During these task conflicts, the participants presented, valued, and argued for their various perspectives, which widened the group’s decision basis. High demands were placed on the participants’ ability to merge their perspectives. Our assumption is that a health care service based on merged perspectives would be more feasible to both lead users and health care providers compared with services developed by more conventional means. However, this assumption must be studied further.

One of the goals for the co-design process was to determine what type of service to recommend for people with RA in order to self-manage PA. The frame, structure, and features to include in the future service were not predetermined, which was frustrating and difficult for the participants. Consistent with previous research, stating clear goals contribute to a more effective collaboration in this study [26]. To handle this challenge, the future target users, the scope, and the aims of the service should be clarified early in the process.

Another challenge in the initial stage of co-design concerned the participants’ roles and responsibilities. The participants were unsure about their respective roles and expectations, which have previously been reported to negatively impact co-design performance [23,27]. Particularly, the academics and professionals had complex roles and were expected to contribute using their unique expertise while considering how to use their competencies without dominating. They were also expected to imagine the needs of the future users. The complex roles for researchers were previously described as a substantial challenge in participatory action research projects [28]. Our study expands that challenge to include not only researchers, but also the other participants in the co-design group. However, the researchers still expressed more concerns regarding their roles and the risk of dominance compared with other participants.

Another challenge concerned the recognition and transformation of the participants’ needs and ideas into concrete features to optimize the future service. The moderator, who had substantial experience with programming and designing digital devices, had a distinct role in this process. Needs and features were defined in terms that enabled the participants with no experience in this field to discuss the features on the future service. The moderator also programmed prototypes that allowed the participants to visualize the future service. The inclusion of engineers to help transform users’ needs into technical applications has been reported previously [27]. Our study emphasizes the importance of including experts in programming and system design to help the co-design participants imagine, discuss, and visualize the features on the future service.

The most important limitation to consider is the fact that the results are based on only 1 co-design process. By involving 4 researchers (ÅR, CM, PÅ, and CK) in the data collection, and 3 in the data analysis, the researchers’ different preconceptions, viewpoints, and analytic skills were used to achieve high credibility of the results. Nevertheless, some important issues may not have been explored.

Another limitation concerned the data-collection methods. We used video recordings and observations to collect data on the co-design process. When planning the workshops, we discussed the possibility of using 2 cameras to enable simultaneous views of the participants, the interactive board, and the moderator, which is technically complicated. Instead, we connected the camera to a computer and used the Wirecast program to enable simultaneous views. As the workshops proceeded, the participants were clearly more active when using plastic sheets and notes compared with the interactive board. Consequently, we lost the possibility of simultaneous views and had to shift the camera focus between participants, the moderator, and the plastic sheets. The 2 naturalistic observers who were present at the workshops provided an additional overview of the process, which complemented the video recordings in a satisfactory way and contributed to data triangulation.

A unique feature of this study was the use of video recordings and naturalistic observations of the co-design workshops to capture the challenges during the process. Previous empirical studies have used diverse methods to describe and explore the co-design process; for example, surveys [4,12,13], observations and audio-recordings [5], and focus group interviews [27]. We have only been able to identify 1 earlier study that used video recordings to study collaboration during co-design meetings [15]. The use of video recordings provided access to detailed data on the interplay of talk, behavior, and context. It also enabled repeated viewings of the video recordings during analysis and gave the authors access to authentic data. The observation protocols facilitated the identification of the challenges and verified the results from the video-based analysis.

A strength was the use of multiple strategies to secure the credibility of the findings. We used multiple data sources, video recordings, observation protocols, and multiple observers. Researcher triangulation was performed during the different stages of data analysis. By providing a thorough description of the setting, scope of the workshops, methods used, and analysis performed, we intend to make it possible for the reader to assess the transferability of the challenges to similar co-design processes [29]. Our co-design process aimed specifically to develop a mobile Internet service for self-management of PA in RA and included multiple stakeholders in 1 co-design group. The transferability might be unique to processes similar in aim and scope. Nevertheless, the identified challenges at a general level may be valid for other co-design processes attempting to develop and improve health care services and should be evaluated in future services.

Participatory action research and video-based analysis raises some ethical issues [19]. The use of video-based analysis revealed each participant’s views, preferences, and actions. Therefore, it is of major concern to thoroughly consider how to present the data to retain confidentiality.

Co-designing the system requirements of a mobile Internet service with multiple stakeholders is a complex and extensive collaborative decision-making process. Considering, valuing, counterbalancing, and integrating different perspectives into agreements and solutions (ie, the merging of participants’ perspectives) were crucial for advancing the process and considered the core challenges of co-design.

This new knowledge of crucial challenges is worth considering when planning and performing future co-design processes on eHealth services that include multiple stakeholders. This study emphasizes that the challenges are crucial for success and should not be omitted but carefully considered and prepared for. Further research is needed to replicate the results in similar and new contexts. Studies on how the participants’ and group characteristics influence the process of merging would also contribute to the field in a significant way. Finally, the inclusion of multiple stakeholders within 1 co-design group is more beneficial than the formation of separate co-design groups for each of the stakeholders with respect to the development and improvement of health care services.