The E = M tool exists of an algorithm, including patient characteristics assessed with a digital questionnaire [i.e., age, gender, PA behavior, Body Mass Index (BMI), medical diagnosis, motivation to change PA and preference to discuss PA with their clinician], set against norm values. The tool and its algorithm are described in more detail elsewhere (22). The online E = M tool provides an individual E = M-prescription for patients and referral options to local PA interventions in- and outside the hospital. An E = M decision guide was developed to support clinicians in the referral of patients to tailored PA interventions (22).

We applied the Implementation Mapping approach, using strong stakeholder engagement in a co-creation process, to develop an implementation protocol to support and guide clinicians during the adoption, implementation and sustainability process of prescription of E = M in a clinical setting. Implementation Mapping is a systematic, multi-step approach for developing implementation strategies that incorporates theory, evidence, and stakeholder perspectives (25) and has been applied previously in our field (27–29).

We operationalized the 5 tasks as follows:

In task 1 – Conduct needs assessment and identifying key actors - we conducted a mixed methods study among clinicians and their managers working at the two participating university hospitals to identify determinants (i.e., barriers and facilitators) for implementing E = M in clinical practice. The needs assessment results are described in more detail elsewhere (18, 22). Therefore, we will only briefly describe these results and how they were used to inform the development process of the implementation protocol. Building on the needs assessment and discussions with the working group of each department and the consortium, we selected key actors to implement E = M in clinical care.

During task 2 - State implementation outcomes - we defined performance objectives for adopting, implementing and maintaining the prescription of E = M in clinical care for each actor. Performance objectives make clear “who has to do what”.

Next in task 3 – Select implementation strategies - we were guided by the Effective Practice and Organisation of Care taxonomy from Powell et al. (23). We used a structured approach by matching our identified barriers and facilitators from Task 1 to the Consolidated Framework for Implementation Research (CFIR) framework classification (30). Waltz et al. have conducted an expert based ranking procedure to identify which implementation strategies would best address specific implementation barriers, resulting in a practical tool for matching implementation strategies (31). This Strategy Matching Tool helps users to choose strategies based on CFIR barriers (30), and complements Fernandez and colleagues' Implementation Mapping (25). Therefore, we linked each identified barrier to a CFIR construct and then selected the top 3 highest ranked proposed strategies by the matching tool.

In task 4 - Produce protocol for implementation and materials - we (FN, JN, AB) matched the implementation strategies (result of task 3) to the suggested practical solutions gathered during the interviews and during discussions with the working group of each department and the stakeholders in our consortium. To operationalize the strategies, we held design sessions with working groups of each department to ask input for the materials. We produced final materials tailored to each department's specific needs and context-specific factors. The tailored implementation strategies were operationalized in a way that delineates what they entail and how they will be delivered and described according to Proctor's reporting guidelines (24).

Lastly in Task 5 – Evaluate implementation outcomes - we developed a mixed methods process evaluation plan for both the processes and outcomes of the set of implementation strategies using the RE-AIM framework (32). The process evaluation was conducted during the pilot implementation in the four departments of two university hospitals (Amsterdam UMC, UMCG), where interviews were initially conducted (33).

During each task of the Implementation Mapping approach, we strongly engaged with stakeholders within our consortium, working groups at the participating departments and patient representatives. The PIE = M project is supported by a large consortium of clinicians working in the participating clinical departments, researchers, information technology professionals, implementation experts, sports organizations, municipalities, lifestyle professionals and patients.

In each department, we installed a working group led by researchers consisting of the manager, clinician, and administrative support. In addition, during six meetings (May, Jul, Sept, Dec 2018, and May, Dec 2019) with the consortium, several aspects of the PIE = M project were discussed and active participation, brainstorms over and reflections on the Implementation Mapping results were gathered.

Although the primary focus of the PIE = M project is on individually tailored prescription of PA by hospital clinicians, a panel discussion with patient representatives was organized to reflect on the developed strategies for implementing E = M from a patient perspective.

The needs assessment comprised of a questionnaire (n = 45 clinicians) and a semi-structured interview (n = 19 clinicians). Our mixed methods needs assessment resulted in 52 identified facilitators and barriers for implementation. A full description of the assessment and its results are described in detail elsewhere (18). Briefly, we identified four main themes: (1) beliefs toward the implementation of E = M (e.g., clinicians knowledge and skills, or social support), (2) factors related to the patient perspective (e.g., patient priorities or motivation), (3) factors related to the E = M referral options (e.g., knowledge of and trust in local referral options) and (4) practical considerations when implementing E = M (e.g., time constraints).

Discussing these determinants with our consortium stakeholders enriched the content and explanation of factors influencing implementation of E = M, but did not lead to new determinants. Yet, the stakeholders emphasized that some determinants were beyond this project's scope, such as knowledge among the general population about the PA guidelines. Consequently, those determinants were not included in the next tasks of the Implementation Mapping approach.

In addition, the stakeholders endorsed that not all determinants will have the same impact at each department. Therefore, tailoring the strategies to the specific departments was mentioned to be essential. We have considered this in Task 4 when designing the implementation materials.

To realize adoption, implementation and sustainability of the E = M in routine clinical care, together with involved stakeholders, we designed objectives for clinicians, department managers and stakeholders in the broader context, by specifying who and what will change due to the implementation strategies. Table 1 provides an overview of implementation outcomes for each actor group to facilitate successful implementation of E = M in routine clinical care.

In Task 3 evidence-based and theoretically sound implementation strategies were selected. Table 2 shows the six themes of determinants (result of Task 1), their equivalent CFIR construct(s), selected implementation strategies based on the Strategy Matching Tool, and timing.

In total, 24 strategies were matched to 32 determinants. We identified 5 strategies for the adoption phase, 6 for the pre-implementation phase, 4 during the start of implementation, 5 for during implementation and 4 strategies aimed at the sustainability phase.

Most often the strategy Identify and prepare champions was matched to address determinants (n = 14, 44%), followed by Conduct local needs assessment (n = 9, 28%). All clusters of strategies were represented, except for the Support & provide assistance clusters.

Guided by Table 2 of Task 3, interview data of Task 1 and by input from our consortium, we translated the theoretical strategies into practical applications included in the implementation protocol (see Table 3).

We operationalized actions needed in the adoption phase to facilitate momentum for the implementation of E = M, such as stimulating discussion, identifying champions, and discussing the added value. During the pre-implementation phase, readiness and needs among clinicians and patients were assessed to tailor the tool for individually tailored prescription of PA for patients to the local context. Furthermore, a training was developed to support clinicians in using the E = M tool, as well as the other strategies. At the start of the implementation process, the training was delivered, all facilities were set to start, and a plan for evaluation was developed.

During implementation, strategies to engage patients were implemented, as well as clinical implementation team meetings to discuss the successes and to address potential barriers.

Lastly, strategies were operationalized to support implementation on the long term (i.e., sustainability) by network weaving, and advocacy for reimbursement by the health insurance company.

Although no strategies of the Support & provide assistance clusters were matched to determinants, we included these in our implementation protocol in the form of appointing a local implementation lead (i.e., formally appoint internal implementation leader responsible for the coordination of the implementation process), as the need for such support was often mentioned during stakeholders interviews and by members of our consortium.

Actors for the designed implementation strategies were mainly the implementation lead, and the managers of implementing departments targeting their actions towards clinicians, managers, patients, other stakeholders involved in the broader context and the board of directors.

Furthermore, our needs assessment highlighted the necessity for lifestyle coaches appointed at the implementing departments and/or hospitals. Lifestyle coaches are health care professionals working in the outpatient clinic, closely collaborating with clinicians. They qualify for lifestyle counselling and give eligible patients one-off consultations aimed at knowledge transfer, improve patients' motivation, and regional referral options.

According to the RE-AIM framework (32), we conducted short cyclical tests to evaluate the feasibility (i.e., reach, effectiveness, adoption, implementation and sustainability) of the implementation protocol in 5 pilots in UMCG between October 2019 and October 2020. A full description of the results is described in detail elsewhere (33). The same departments were also participating departments in the Implementation Mapping tasks, therefore, many adoption strategies were not evaluated during the pilot phase, as they were already applied during the co-creation phase. After the co-creation process, 5 strategies were piloted in the Departments of Orthopedics and Rehabilitation Medicine of UMCG, the Netherlands. The implementation lead was a researcher (AB), who was responsible for (1) educating clinicians on E = M, (2) incorporating the E = M tool in the Electronic Medical Records, (3) appointing of lifestyle coaches, (4) generating an overview of local E = M referral options, and (5) providing support to implementing teams. Through short cyclical evaluation, each implementation informed adaptations that were put into practice in the next implementation cycle. In total, 12 clinicians of 2 departments participated in 5 pilots. During the pilots, 210 patients were invited for participation and clinicians reported that they discussed the tailored advice with their patients in 47% (Orthopedics) to 62% (Rehabilitation Medicine) of their consultations.

Interview data of the process evaluation showed that clinicians particularly valued three of the five main implementation strategies: namely education on E = M, the lifestyle coach appointed within the department, and the presence of an implementation lead. While the E = M tool in the Electronic Medical Records was appreciated, clinicians identified several barriers that hindered its effective use in practice. Additionally, the overview of local E = M providers was rarely used by clinicians. Based on the evaluation, deployment of a lifestyle coach within a department, and implementation lead deemed essential. Clinicians expressed a need to structurally embed lifestyle coaches in the hospital, with especially knowledge about local E = M providers.

This study has several strengths. First, the study follows a systematic approach to develop an implementation protocol, ensuring that the selected strategies are grounded in theory and practice. A second strength is the use of a matching tool, which has been central to the study's methodology. Given the current stage of the field, this tool remains the best available method for matching strategies to determinants. Lastly, multiple data sources were used, including interviews, input from different departments, and insights gained during consortium meetings, providing a rich and diverse perspective to inform the implementation protocol.

Several limitations need to be considered. First, the generalizability of the findings to other departments or hospitals is limited, as the study primarily focused on the Dutch academic clinical setting, and departments that already focused on PA. Yet, the already matched list of potential strategies described in our manuscript in Table 2 can serve as a starting point for selection of strategies, that match local barriers. Other hospitals can then adapt these to their local contexts, and use our RE-AIM evaluation protocol to evaluate implementation processes.

Another limitation lies in the process of matching our barriers and facilitators to CFIR constructs. Many of the terms used by our stakeholders (e.g., during interviews or consortium meetings) are framed in practical, everyday language, which we may have misclassified to CFIR constructs. However, during the operationalization, we aligned all tasks with our consortium members and patient representatives to avoid a mismatch of our strategies.

This study illustrates the application of Implementation Mapping to design an implementation protocol to support and guide prescription of E = M by clinicians, using strong stakeholder participation in the development process. Implementation Mapping proved to help develop the protocol with our stakeholders. As such our stepwise development of the implementation protocol can serve as an example for researchers or practitioners preparing for E = M implementation.