The concept of boundary objects was introduced in 1989 by Star [1] to describe how specific artifacts can fulfill a bridging function between different sociocultural sites. Over the past decades, there has been more interest in boundaries, boundary crossing, and boundary objects [2-6]. The idea of boundary objects was initially framed to facilitate constructive cooperation between sites or social systems without consensus [7]. This organizational feature of boundary objects can be of great value in health care innovation, where the promise of innovations outweighs their actual impact [8-13]. Owing to many different stakeholders and parties in health care with different needs and goals, the implementation of innovations in health care practice is complex [14]. Many frameworks on innovation and implementation pursue consensus [15-21], mainly from a monodisciplinary approach, where at some point, all parties and stakeholders must be convinced that an innovation is of added value from a specific viewpoint. Within disciplines, this can be feasible, but across disciplines, this is often challenging. Boundary objects offer a different perspective on this issue. Boundary objects ideally address the needs of each stakeholder group and aim to contribute to the goals of all stakeholders involved, even if they do not pursue the same goal. This also means that different stakeholders can interpret a boundary object differently, something that Star [7] calls interpretive flexibility. However, not striving for consensus but identifying and addressing needs on the front end requires a fundamentally different approach. Much more attention will need to be paid to the preimplementation phase, which is seldom included in frameworks [22]. The design discipline is a discipline par excellence that pays attention to precisely this phase.

Design research and design thinking are increasingly finding their way into the health care sector as appropriate methodologies of responding to a world with more open, complex, and increasingly networked problems. Design holds the promise of offering suitable strategies for complex problems and actively involves stakeholders during the development and implementation of innovation [23,24]. Design as a discipline already has a long history in the development of medical devices but is now broadening its scope in shaping the future of health care [25-28]. Owing to different causes, the worlds of health care and design are converging. In health care, there is a shift in focus toward patient experience and values, increasing the quality of life and patients’ participation in care and treatment [29,30]. In the design discipline, developments toward phenomena such as experience design [31], value-sensitive design [32], and people’s involvement in design through participatory design [33-35] seem to have a good fit with the shifts in the focus of health care. The focus of emerging design disciplines on innovation, transformation, and services within organizations [36] can also solve implementation and adoption problems in innovation. However, many frameworks or models that provide insight into shaping the design process focus more specifically on the steps, methods, or guide points essential for developing an artifact [37-42] and less on shaping the process of change.

The following six databases were searched for potentially relevant abstracts: PubMed, Scopus, Education Resources Information Center, PsycINFO, Information Science and Technology Abstracts, and Embase. These databases cover a wide range of published research in the field of health care. They were selected after several trial searches in various databases and after consultation with an information specialist in health science. The terms that were used for the search in PubMed are presented in Textbox 1.

Owing to differences in search engine functionality, the method by which terms were entered differed per database. A complete overview of the terms is included in Multimedia Appendix 1. Searches included papers published between 1989, when Star [1] introduced the concept of boundary objects, and September 2020. Before the definitive search, we performed 3 trial searches with different terms to reduce the possibility of missing relevant studies. We conducted a definitive search on September 23, 2020. We followed the PRISMA (Preferred Reporting Items of Systematic Reviews and Meta-Analyses) guidelines [78] as much as possible to report this review.

We included studies that discussed boundary objects or innovations in health. We included only original reports or papers that (1) mentioned boundary objects, (2) involved an empirical study, or (3) otherwise focused on a newly developed or implemented innovation. Papers meeting these criteria were selected for full-text screening.

The following exclusion criteria were used for full-text screening: (1) non–peer-reviewed papers such as abstracts, conference posters, or trade journals; (2) papers with full text not available; (3) papers in languages other than English; (4) monographs or short reports; and (5) papers with not sufficient information in the abstract.

After removing the duplicates, the papers were screened based on title and abstract using Rayyan [79]. A total of 2 reviewers (GT and DK) independently reviewed all titles and abstracts, who were double-blinded for relevance with the formulated inclusion and exclusion criteria. Papers were only included on the agreement of both GT and DK, where a plausible argumentation for consideration of inclusion always led to inclusion. Full-text papers were retrieved after this step. During the full-text screening phase, the first 20% of the papers were randomly selected and double-blind reviewed by 2 reviewers (GT and LV). After this scan, no disagreements occurred about inclusion or the identified mechanisms. Then, the main reviewer (GT) reviewed the other included papers for a full-text reading.

Our initial search yielded 3102 records. Of the 3102 records, after removing the 916 (29.53%) duplicates, 2186 (70.47%) records were screened based on their titles and abstracts. Next, of the 65 records, we screened the full text, leaving 25 (38%) papers for inclusion (see Figure 1 for a flowchart of the results in the different selection stages). In both stages, there was a consensus between the reviewers on both the inclusion and analysis of the papers.

The studies included in this systematic review had varied study designs and topics. Table 1 presents the study designs, topics, and characteristics. All the included articles were published after 2008.

We studied full-text papers on applying the concept of boundary objects; whether this concept was used to describe daily life situations or situations where there was innovation either in its development, implementation, or postimplementation stage; and if ≥1 dialogical learning mechanisms could be identified. In the following section, we categorize the papers based on their mechanisms and the situations they applied to.

Of the 25 papers, 2 (8%) [81,93] described a medical term as a boundary object to identify different medical terms’ interpretations over different contexts or disciplines. Lambert et al [81] approached terms such as infection, antibiotics, and inflammation as boundary objects. Williams et al [93] conceptualized embryos to find out how they were decontextualized and recontextualized within and between 2 different cultural systems. In both studies, existing beliefs were reinforced, and the studies aimed to identify local differences by applying the identification mechanism in a noninnovation context.

In 8% (2/25) of the papers, the identification mechanism was deduced in more innovative projects. Nielsen and Mengiste [80] described in their study an analysis of the adoption of mobile information technology innovation for home care. The innovation functioned as a boundary object on the level of influential stakeholders (ministry, local government, and managers). Managers and care workers revealed different interpretations of the technology’s value and potential, resulting in resistance and tension. The technology seemed to reinforce existing differences between activity systems at the level of managers and care workers. The authors proposed a bottom-up approach and more involvement of end users in the future. This technology seemed to reinforce existing differences, triggering the identification mechanism.

Bjørn et al [83] described “conflicting perspectives between standardization and reconfiguration embedded within hospital information systems (HIS) design activities.” In their study, the authors considered an electronic triage and tracking system as a boundary object. Users indicated whether they could work with the system and how. This led to adjustments and reconfigurations in the system and, presumably, better adoption of the system but not to, for example, perspective taking between different groups. Differences were primarily sought for user groups to use the system optimally by having them respond to the system, thus using the identification mechanism in this study in a constructive way to retrieve specific input.

In 16% (4/25) of the papers, we identified the coordination mechanism to facilitate cooperation in daily practice. Approximately 50% (2/4) of these papers presented controlled clinical vocabulary to facilitate and coordinate collaboration between different professionals [86,87]. Approximately 50% (2/4) other studies make use of data or narratives to facilitate multidisciplinary cooperation. Meier [92] described an ethnographic study of 2 hospital wards. Patients’ stories, especially their narratives and patient records, formed the boundary objects to make constructive multidisciplinary work possible. A study by Isah and Bystroöm [97] focused on the role of case notes as mediating artifacts in patient care. Their study demonstrated how case notes were a source of information and an essential, enacting, and mediating part of the work itself. The case notes seemed to be multipurpose; they served as a repository of information and knowledge and supported and mediated a plethora of the medical team’s work activities in patient care. It was evident that the case notes served as a coordinating mechanism between the participating actors. Besides facilitating and fortifying many day-to-day functions in patient care, case notes have established themselves for deliberate learning; they embody a clerkship template and enable newcomers to integrate and perpetuate the practice.

In 12% (3/25) of the papers, we identified a coordination mechanism to promote innovation. The coordination mechanism was identified in the first of 2 included papers from Sajtos et al [91], where the authors reported a case study of developing a case-mix system. Their study illustrated a process to address the diverse meanings and interests of various stakeholders to overcome communication and organizational challenges. They presented a funneling framework. A so-called boundary concept evolved through stakeholder input into a boundary object in a second step and a solution in the final step. Both principles and constraints were identified and addressed in the final solution by aligning the stakeholders’ interests. In the project, no clients were directly involved, and it remains unclear whether and how the design itself was subject to flexibility along the way or that the project was mainly about fine-tuning a design.

Gregory et al [98] evaluated a diabetes eHealth system in their study. They described the use of the system as a boundary object for developing an understanding of why the eHealth system was used in a wide variety of ways, enabling coordination over stakeholder groups. McLoughlin et al [102] reported 4 case studies of health information infrastructure projects, where they approached the health information systems as boundary objects. Two of the projects were on a regional level, and 2 projects were on a national level. In the regional projects, the boundary objects managed and facilitated collaboration in using health data for different purposes by different users. At the national level, the 2 boundary objects instantiated top-down attempts and struggled more to trigger some effects.

We identified no papers which merely reported the reflection mechanism.

The transformation mechanism was identified in 8% (2/25) of the studies. Both studies described a more extended transformational development in retrospect. Fox [88] described the development of antiseptic and aseptic environments during surgery. In a historical case study, the report assessed the innovations in surgical sterility and how boundary objects worked over time. For example, in this study, nose masks were considered boundary objects in their relationship to social meanings within communities of practice. In the conclusions, the researcher described positive and negative boundary objects and concluded the following:

This study described the surgery profession’s transformation process, partly through boundary objects, from only a healer of disease to a healer of disease who was also a safety procurer.

The second historical perspective was written by Stewart and Watson [99], who described the development of the ultralightweight wheelchair and its social implications. As a boundary object, the ultralightweight wheelchair had a significant transformational impact on the use of wheelchairs in the daily lives of users of wheelchairs. According to the authors, the wheelchair as a boundary object provided many insights through various interpretations of the artifact. It reflected views about users of wheelchairs and disability more generally and how the ultralightweight wheelchair as a boundary object seemed to manifest power relations between the diverse communities it engaged.

This review shows that the concept of boundary objects has found its way into health care. The use of the concept has been growing since 2008, with a significant number of papers describing boundary objects from the past 5 years. In the reviewed studies, we see that boundary objects are mainly used to shape and organize multidisciplinary work, close to the original explanation of Star and Griesemer [46], or to surface differences in, for example, interpretation of a concept from different contexts or disciplines. In the 25 papers, 38 mechanisms were identified, of which 15 (39%) were coordination mechanisms, and 10 (26%) were identification mechanisms. In addition to the organizing and performative effect, boundary objects can reinforce boundaries and create conflicts. In addition to the proposition by Star and Griesemer [46], Oswick and Robertson [103] referred to barricades and mazes that generate conflict and reinforce boundaries and existing differences, something that Langley et al [48] also described as part of competitive boundary work. This can be an opposing and perhaps unwelcome side of the identification mechanism in terms of change management. In the study by Kajamaa [82], we saw this effect. First, in what seems a fluid development and implementation process, they applied the identification mechanism to identify different stakeholders’ needs. After implementation, 1 event led to a breakdown of trust between stakeholders, which led to the project’s withdrawal. After this event, the boundary object was primarily used to name the significant differences between stakeholders and compete for a position without the other.

Although the concept of a boundary object was introduced to describe how specific artifacts can fulfill a bridging function between different sociocultural sites [1] and thus, have a social focus, the focus in the included papers was often on the boundary object itself rather than the social effect. Various labels were given to boundary objects in different studies, which described a more designerly process for an artifact. In the different included and excluded studies, we saw a differentiation between designated boundary objects and boundary objects in use [96,104,105]. This differentiation can be seen as parallel to the design research process. An artifact or solution continues to take shape and is developed in small steps from prototype to object in use. The analogy can also be made by applying a boundary object from a more ill-structured to a more well-structured context. In a second included paper of Islind et al [44], they described three types of boundary objects: narratives as open boundary objects in the first phase, metaphorical boundary objects as semiopen boundary objects in the second phase, and structured boundary objects in the third phase. Although again, the focus was more on the development of the object itself rather than the effects of the object in the social context, parallels can be drawn with a design process and application of the learning mechanisms in practice, as can be seen in the results.

The reflection and transformation mechanisms are underrepresented in the included studies. Of the 25 studies, 2 (8%) describe the transformative effect of boundary objects from a historical perspective [88,99], describing a long timeline of a particular development. However, it is difficult to determine the impact of the boundary object itself in retrospect as it is likely that many more variables played a role in the transformational processes. In addition, it is difficult to determine, in retrospect, whether the boundary objects were deliberately deployed for the given purpose. The reflective mechanism was the least identified in all the papers. However, in the papers in which the reflective mechanism took place [57,82,84,85,90], there was a much smoother adaptation and application of the innovation or tool afterward. There was more shared ownership of the problem and solution in the processes described and more consideration of other perspectives along the way. This reinforces the idea of reflection as an essential step in the design process, especially in a more complex setting with multiple stakeholders, needs, and interests. When these are appropriately addressed in the design through a boundary object’s focus and, simultaneously, addressed within the design, more mutual understanding arises. This leads to a natural emerging change space where everyone is willing to move forward [76,77].

On the basis of the findings of this study, for future design and implementation projects, the social focus of boundary objects can add value to innovation projects. Pursuing to trigger the reflective mechanism can lead to the benefit of more fluid and smooth integration of innovation into practice. Here, the boundary object perspective avoids the pursuit of consensus, which often proves unfeasible in complex practices with many stakeholders. The reflective mechanism creates a shared awareness that there are multiple perspectives and needs. This awareness can lead to a shared change space in which innovation can flourish.

As seen in previous research [106], little attention has been paid to describing a conscious rationale for designing innovative artifacts in health care research. This makes it hard to determine the thoughts and foundations of a designed object. In this study, this fact also made it difficult to ascertain the intent behind the deployment of particular boundary objects. The effect was often identifiable; however, it was impossible to determine whether it was directed or accidental without knowing the intention. In addition, no study described what changes were explicitly made to a prototype or design after a specific stakeholder workshop or meeting. The often implicit focus on effect is evident in health care research, making it difficult for innovative design processes to get sensible insight into the design rationales of others.

Another observation was that many innovations in the included studies were more administrative systems, such as electronic patient files. These are pre-eminently systems with which different disciplines must work, and boundary objects are thus helpful; however, 12% (3/25) of studies showed that boundary objects are also of added value in research in which clients or patients have an active role. This observation raises the idea that there are still more gains that can be found by involving end users earlier in design processes.

The included papers were subject to the interpretation, discussion, and consensus of the reviewers (GT, DK, and LV). To counteract subjectivity as much as possible, papers were double-blind reviewed by 2 reviewers in the title and abstract scan (GT and DK). They were only included in the consensus of both reviewers. In the full-read phase, 20% (13/65) of the papers were double-blind reviewed by 2 reviewers (GT and LV) before they were discussed. No disagreements on inclusion occurred during the discussion.

The concept of boundary objects has found its way into health care. In this review, we saw that boundary objects in health are primarily used to shape and organize multidisciplinary work or to surface differences in, for example, the interpretation of a concept from different contexts or disciplines. Although the concept of a boundary object was introduced to describe how specific artifacts can fulfill a bridging function between different sociocultural sites and thus have a social focus, the focus in the included papers was often on the boundary object itself rather than the social effect. The reflection and transformation mechanisms were underrepresented in the included studies; however, based on the findings in this review, pursuing to trigger the reflective mechanism in design, development, and implementation projects can lead to the benefit of more fluid and smooth integration of innovation into practice.