This formative participatory action research (PAR) study followed 5 clusters of health care professionals over 6 months using 3 interviews per cluster and a model to guide an iterative inductive thematic analysis.

PAR is a collective, self-reflective investigation undertaken by researchers and participants together [42]. It connects actions influenced by context, culture, and history and is embedded in social dynamics. The strengths of PAR are responsiveness to context, the engagement of frontline health care professionals, and a focus on the mechanisms of implementation that can help bring about real-world service improvements [43].

Throughout the successive meetings with participants, we followed the iterative process proposed by Loewenson et al [44], using the steps of systematizing experience, collectively analyzing and problematizing, reflecting on and choosing an action, taking and evaluating action, and systematizing learning. We invited each group of participants to define their collective commitments at the first meeting. The reflective process was stimulated by asking questions such as “What is going on?” “How do we continue?” and “What are our main lessons learned?” We also ensured that all the lessons learned that were documented by researchers were proposed for further discussion or refinement.

The study was embedded in the pilot project for the implementation of shared electronic medication plans on the regional eHealth platform for the Nord Broye region in the canton of Vaud [45]. We recruited health care professionals into local study clusters from among the 36 general practitioners (GPs) and 36 pharmacies who had cared for the 193 patients participating in the pilot project from 2013 to 2018 (Figure 1). Primary care professionals were free to enroll in the pilot project led by the regional network for care coordination, which was sponsored by their respective corporation and public authority. Patients using at least three medications regularly were invited to join the pilot project’s medication management program. Care professionals communicated to patients directly, while the pilot project team provided leaflets and information online. They nominated a GP and a pharmacy as reference points to manage their medications, and they committed to consulting and procuring their medication only from them while sharing all necessary information completely.

The digital solution chosen for the shared medication plans was an online platform for creating, using, and managing a list of all the medications a patient was taking and had taken in the past [22,46]. A shared electronic medication plan must be accessible, complete, and updated at every contact between the patient and an intervening health care professional. Technologically, this solution was envisioned as an interoperable system based on the pharmacy profiles defined by the Integrating the Healthcare Enterprise (IHE) consortium [47]. During the pilot project, the definition of a national e-medication interoperability standard based on IHE pharmacy HL7 CDA was under development (interprofessional working group from 2015, recommendation published in 2017) [22,48]. The users accessed the shared medication plan through a secure web portal with two-factor authentication. Patients could access their medication plan online or receive a printed one. Professionals had to enter all data manually in addition to filling out the usual paper documentation because their clinical software applications were not yet integrated.

The solution is a module of the web platform developed for the cantons of Geneva and Vaud in anticipation of Switzerland’s electronic patient record (EPR) system, a national digital inventory of all the relevant health data concerning the country’s patients [22]. The EPR is based on decentralized information exchange infrastructure. Several regional platforms have been implemented that are run by private or public entities but overseen nationally by the federal law of 2017 [49]. Patients own their data and share them with health care professionals of their own free will. Primary care physicians are free to choose whether they want to join the EPR (opt in), whereas all hospitals are obliged to be connected. Swiss national policy acknowledges the importance of e-medication [50,51], but an overall strategy has not yet been defined.

Switzerland’s political culture is liberal, and the 26 cantons of the federal state have far-reaching autonomy regarding the organization of health care [52]. Patients can access the health care professionals or specialist physicians of their choice. GPs only have a gatekeeper function in some optional insurance plans. As of 2020, there are no shared patient registers. Among the member states of the Organisation for Economic Co-operation and Development (OECD), Switzerland has below-average digital maturity [53], and 30% of its GPs still use paper-based patient records, far behind their colleagues from the European Union, of whom only 4% rely on paper. Finally, there are no regulatory or other specific incentives for the vendors of medical or pharmaceutical record software.

Organizational innovations in health care often fail because the complexity and adaptability of the health system are underestimated [54-56]. Recognizing health care as a complex adaptive system (CAS) means focusing on the dynamic interactions between individuals and organizations across the entire system. When seeking to initiate change in a CAS, sensemaking and learning about it are more important than planning and controlling the change itself. By definition, a CAS is unpredictable, but some simple rules (ie, guiding principles), can help foster transformation [55,57-59].

Creating a common electronic patient health record incorporating a shared medication plan is in itself a complex sociotechnical intervention; the technological component of the intervention is influenced by and influences every user’s behavior, as well as the organization and context [60]. We developed a model (Figure 2) of how different elements of the implementation of the shared electronic medication plan might be linked to expected results.

We based our model on Lilford et al’s [61] approach to mapping policy and service interventions with regard to structures, processes, outcomes, and intervening variables. In the pilot study, the introduction of the shared electronic medication plan system affected the participating health care organization structures and required adaptions of their work processes. Some participants also combined the shared medication plan use with other clinical interventions, such as medication review. All these elements as a whole system led to health care outcomes. Although we did not want to predefine the intended outcomes of safer and better patient care, we did specify that the continuity of care, claimed as a main policy ambition for the pilot project, should be not only at the informational level but also at the relational and management levels, as per Haggerty et al’s [62] definition. It is also essential to consider the intervening variables, as they are interrelated with the structural and process factors mediating the outcomes. For instance, a patient’s trust in their pharmacy and its staff (intervening variable) is influenced by the availability of a space in the pharmacy where they can talk in confidence (structure), whether a dedicated pharmacist follows up with a chronic patient (generic process), how information is given when dispensing a pillbox (clinical process), and the consequent safe use of medicines (outcomes). In the present study, the main intervention is at the policy level: implementing shared electronic medication plans on the eHealth platform in the region. Our study sought to leverage health care professionals’ experiences to assess contextual influences from a systemic perspective. For this reason, the model specifies both the context and the readiness of the provider or the patient, as structural factors can be respectively external or internal of the health care providers.

The information system itself was added to the model as a transversal dimension, based on the eHealth Clinical Adoption framework defined by Lau et al [63]. Those authors described how the successful adoption and benefits of HIT depend on its quality. The overall quality of HIT is made up of the qualities of the system, the service, and the information available. For a shared record system, because the quality of information is made up of shared content, it is strongly dependent on the quality of usage. This is why our model illustrates the interrelation of the perceived quality of the shared electronic medication plan system, the quality of its usage, and the quality of the shared content as distinct dimensions. Finally, the model describes the shared electronic medication plan’s overall added value in terms of the improved elements in the continuity of care and the benefits of HIT.

All the dimensions in our model helped us to break down and make sense of the implementation of the HIT, potential interventions, and the points requiring study. The two essential new elements brought in by the pilot project were the addition of a shared electronic medication plan system onto an existing eHealth platform and a new collaborative model based on the patient-GP-pharmacy triad to make primary care medication management safer in cases involving polypharmacy.

Invitations were sent to the group of 36 GPs and 36 pharmacies who had been enrolled in the pilot project. From those who volunteered, we created clusters consisting of at least one pharmacist from an enrolled pharmacy and one GP who were responsible for at least one common patient enrolled in the pilot project. Each cluster could invite other primary health care professionals involved in their local settings, such as a home care nurse. We characterized each cluster by urbanization density classification [64]. All participants consented according to the canton of Vaud’s legal, privacy, and ethical requirements. No patient data were collected.

Two researchers, a pharmacist and GP respectively, both with research training and experience, collected and analyzed the data (Figure 3). They were not enrolled in the pilot project (ie, did not count in clusters) but were familiar with the settings, and they knew some of the participants professionally.

Between May 2018 and January 2019, each cluster participated in 3 group interviews whose main topics of investigation were, respectively, (1) motivations and commitment, (2) experience and refinement, and (3) synthesis and learnings. Interview guides (Multimedia Appendix 1) for each round were prepared using the conceptual model as a basis. The investigators guided participants toward thinking about the added value of the eHealth platform and the collaborative model of care as 2 interdependent components associated with the implementation of patients’ shared medication plans. Participants were first encouraged to share and reflect on their experiences of initiating and managing the medication plan, using the platform, interacting with patients about medication lists, and collaborating with other professionals. These experiences then nurtured discussions on the contextual or organizational factors influencing implementation and on the role of shared electronic medication plans in achieving safer, more effective care. The investigators facilitated exploration of the different themes that emerged from each group in order to increase diversity across clusters. During the last cluster meeting, participants also discussed and summarized the most important practical knowledge that should be disseminated to stakeholders involved in the future development and scale-up of shared electronic medication plan systems.

Data collection and analysis were iterative so that each group’s experiences could be collected and synthesized longitudinally and data across clusters could be analyzed horizontally to condense them into themes. Finally, we conducted a secondary thematic content analysis to condense the themes into lessons learned, or lessons intended to describe the simple rules underlying the mechanisms related to the implementation in a CAS.

Following the principles of PAR, we proposed refinements to and requested validation from participants at each step of the study. Furthermore, we presented the lessons learned at a stakeholders meeting, which included participants and representatives of other stakeholders in the regional pilot project. This gave time for discussions and dialogue on setting prioritized next steps. We also attempted to enhance the reliability of our research by having data analyzed by the 2 main researchers and then by a researcher outside the pilot project [65].

Among the 36 GPs and 36 pharmacies who were enrolled in the pilot project, 31 volunteered for this action research study. A total of 5 clusters were identified, including 13 care professionals. Consequently, 18 volunteers had no patients in common with any other volunteer professionals and thus were excluded in this study. The 13 participants invited 5 extra primary health care professionals into their clusters, for a total of 18 participants (Table 1). We conducted 15 group interviews that lasted 60 to 105 minutes.

An overview of the lessons learned is presented in Textbox 1.

Health care professionals and patients alike need an accessible, common, complete, and accurate list of all the medications the patient is taking. However, introducing shared medication plans has proven difficult in several countries, and guidance for their implementation seems needed. We have presented 10 lessons learned from the first pilot project in Switzerland attempting to implement shared electronic medication plans, and we discuss this in light of studies from other contexts.

Clearly, no single organization can create and implement a comprehensive, robust, and user-friendly shared electronic medication plan system alone; HIT companies, policy makers, project teams, and the system’s users—both professionals and patients—must also collaborate. Given the systemic and safety implications of implementing eHealth projects, public health authorities are taking significant steps to improve the usability of HIT systems [66]. The pilot project suffered from a lack of cooperation among HIT, clinical, and policy stakeholders and from weak enabling framework conditions, especially at the federal level. These external issues prolonged the project phase, contributed to the lack of evolution of the eHealth platform and the absence of integration with other HIT applications, and ultimately led to disengagement by health care professionals. Usability “does not heal by itself” [67] through market competition. Federating the stakeholders in an appropriate, adaptable framework involving collaboration and policy coordination is a sine qua non for the successful implementation of ambitious eHealth projects. Building a shared electronic medication plan system implies a shared ownership.

Implementing a shared electronic medication plan system and improving clinical practice is a complex process. Stakeholders face a dilemma. On the one hand, better clinical practice requires change, which technology can support. On the other hand, the new technology needs to be fitted to an existing process to increase its acceptance. Study participants emphasized that improvements required good clinical practice, trust, and collaboration. Technology alone, therefore, is clearly neither a prerequisite nor a guarantee for safer work processes; rather, it acts as a catalyst [60] for the simultaneous innovation of the technology, processes, and relationships [68,69]. eHealth platforms could be better implemented by using approaches from quality improvement [70] and service design [71]. Combining HIT system design and clinical practice improvement within a shared electronic medication plan’s implementation strategy could likely prevent the chicken-or-the-egg dilemma and better leverage synergies.

A shared electronic medication plan should improve coordination in variable and changing contexts by using the same common regional HIT system. We accepted any proposition from the participants and found variability in how the shared electronic medication plan was initiated and updated across the 5 clusters of health care professionals. In the clusters, the main professional who regularly reviewed and updated the shared electronic medication plan was different; in 1 cluster it was a pharmacist, in 2 it was a GP, and in 2 it was a nurse or the medical secretary with a care coordination role in the GP practice. The basic rule was that they needed to define how they would manage the shared medication plan together in routine practice to ensure its accuracy. The model was easily adopted in every case because it was based on a consensus and the professionals’ preferences on how they wanted to manage it. Participants acknowledged their inherent shared ownership of the shared electronic medication plan. At the same time, clear processes and responsibilities are called for, both in HIT design and among health care professionals [28,32]. Our findings suggested that there was no one-size-fits-all solution; thus, strictly enforcing the implementation of a rigid solution could be difficult and could cause unintended consequences, and it would be unlikely to be achieved through policy making in Switzerland’s context. A strategy enabling all health care professionals to be involved in a patient’s care via a shared system and promoting basic principles of use seems more appropriate. Such an approach facilitates regular updates directly when interventions are made or discrepancies are identified. It may also increase the sense of shared ownership and favor self-organization at the local or the patient level. Knowing the issues related to the complex workflows that hinder the implementation of MedRec [17], an eHealth platform will not likely solve every problem. Standardization, automation, user constraints, and clear roles and processes [28-31] all need to be carefully balanced, with room for adaptations to local variables, in order to support a mutual commitment to using patients’ common medication plans along the continuum of care.

The process of managing a shared electronic medication plan also raises questions about patients’ roles and responsibilities. The pilot project HIT system implemented in the present study was not designed to empower patients and facilitate their engagement in their own medication management. The only function available to patients via the patient portal was a view of their medication plan. They could not make adjustments. The limitations of such an approach are obvious: safe, efficient medication management requires contributions from all stakeholders, including patients and their relatives. Despite increasing evidence of the benefits of comanaging digital medication systems with patients [72-75], they are still mostly treated as the passive recipients of medication lists produced by and for health care professionals. The German experience is insightful. Despite a clear policy for the systematic production of medication plans by professionals, expectations have fallen short. Few of the eligible patients ended up with the accurate list they were supposed to have [76], and when they did, only about half of them understood its content [77]. We plan to further study patients’ perspectives in our ensuing work. Today, any service improvements or innovation should acknowledge the coproduction of value by patients and health care professionals together as partners [78].

Guiding stakeholders’ actions towards the meaningful use of shared electronic medication plans should start by acknowledging the shared ownership and complexity of the process. From a CAS perspective, a strategy for driving major changes relies on the power of an attractor [79], a vision shared among stakeholders, that can inspire independent people and organizations to self-organize and evolve in a coherent, synergistic manner within the broader health care system. Advocating for a shared electronic medication plan comanaged by patients and health care professionals as a shared vision is even more important in settings where stakeholder fragmentation and autonomy are high. Our study stimulated collaborative actions by raising awareness of the value and shared ownership of a shared electronic medication plan, which encourages leadership at every level and supports collective learning. Indeed, these are some of the key ingredients for successfully enabling transformation in a CAS [57,80,81].

Mobilizing stakeholders through formative action research is a promising approach to dealing with the complex sociotechnical challenges related to shared electronic medication plans. This type of research can nurture the implementation dynamic; policymakers cannot mandate the required motivation and trust. Local networks and cultures can vary and have a significant influence on whether a new shared HIT system gains acceptance. Disregarding them has contributed to ineffective communication with the public or failure to engage with health care professionals [38,82]. Health care professionals want to be considered long-term partners in major HIT projects, not simply clients [83]. The series of cluster meetings during our study helped to enhance mutual understanding, collective learning, and trust. Similar benefits have been reported from facilitating an interprofessional dynamic [33,84], especially when it was a core focus of the implementation strategy [29,34]. Our study participants rarely have opportunities for dialogue and reflection at the local level, and this was appreciated and even triggered some further collaborations. We also realized that the mixed status of our 2 main investigators, who were clinicians, researchers, and employees of the public health authorities, strengthened our participants’ motivation to get involved in the study. They considered involvement to be a meaningful way to facilitate communication and mutual understanding between the people in the field and decision makers. We argue that further formative action research could be a key facilitator in the implementation of new shared electronic medication plan systems.

Our study has some limitations. First, we only included GPs, pharmacists, and nurses involved in primary care. Thus, we could only investigate issues related to care transitions from the primary care perspective. Second, we report experiences from a relatively small region of French-speaking Switzerland, which might limit the study’s transferability to other contexts. The lessons learned could, nevertheless, support learning in other settings. Third, the participants who volunteered for the study were likely early adopters and highly motivated. Additionally, because implementation intensity was low, some specific use and implementation issues that are likely to be encountered in the future require more assessment.

However, our novel approach, which used 5 clusters and iterative participatory analysis, is a strength of our study. We maximized diversity by including rural and urban settings, whereas earlier studies were mostly limited to university medical centers. Early adopters are not the majority, of course, but they are often the determinants for the diffusion of any innovation. Leveraging their experience can benefit other individuals less keen to explore that innovation. Moreover, we sought to embrace complexity by using a systems perspective to support sensemaking and awareness. These can help guide stakeholders and likely support further learnings.

The 10 lessons learned from this study give an overview of the mechanisms and dimensions related to the implementation of shared electronic medication plans in primary care settings. This paper gives practical guidance on implementation and describes some of the key sociotechnical challenges that will face implementors aiming to instill the regular, meaningful use of shared electronic medication plans—plans that should be consistently reconciled along the patient’s entire care pathway—in clinical practice.

We consider the poor spontaneous patient involvement with their shared electronic medication plan to be a significant shortcoming and a point that has clearly not met the policy ambitions of fostering patient empowerment and medication adherence. Nevertheless, the local adaptability of the participating clusters was striking, as was their ability to reach consensus around useful solutions. This suggests that implementation strategies should facilitate the emergence of local engagement rather than implementing rigid top-down processes. HIT systems should be able to support various configurations of use in practice while maintaining predefined basic principles agreed among stakeholders. Last but not least, collective leadership is essential to handle the inherent shared ownership of a medication plan and to make change happen at every level, from direct patient care to the policy framework.

Future research should explore experiences in different countries in order to determine how system characteristics, stakeholder cooperation, health care policy, patients’ and professionals’ responsibilities, and implementation strategies affect the uptake of such shared systems by health care professionals and the benefits these shared medication plans bring to patients and health care services overall. Integrating patients so that they begin to comanage their medication plans also raises important questions. Finally, we suggest that formative participatory action research, including qualitative and quantitative methodologies, should play a key facilitating role in achieving a safe and meaningful use of shared electronic medication plans to create an efficient learning health system.