A qualitative approach has been applied to understand the HCPs’ expectations before implementation and their experiences of using the system. Individual interviews and focus group interviews have been used to obtain a rich understanding of the implementation-related issues. Due to the limited number of users of the mHealth app, a qualitative approach was deemed suitable to gain an in-depth understanding of the factors causing the failed implementation. The Consolidated Framework for Implementation Research (CFIR) by Damschroder et al [23] was used to guide our analysis.

The study was conducted in a public municipality home care organization (MHCO) and the home care sections of 4 public academic primary care centers (PCCs) in a rural part of Region Stockholm.

The home care organization provides medical and supportive care for a diverse range of patients, predominantly older adults but also younger individuals with medical needs. The MHCO has the main responsibility for home care, whereas the PCCs provide some home care, for example, the first 2 weeks after discharge from the hospital and the support and service for persons covered under the Swedish Act Concerning Support and Service for Persons with Certain Functional Impairments. At the time of this study, there was no interaction between the PCCs and the MHCO.

The PCCs all use the same EHR system, accessible through the mHealth app, whereas the MHCO has a different EHR system, lacking accessibility through the mHealth app. Figure 1 presents an overview of the study setting.

The intervention consists of a tablet computer with several apps (eg, decision support, online prescription tools, and online supplies ordering), and the new mHealth app that provides mobile access to the EHR (example screenshots from the app are in Multimedia Appendix 1). In the following text, we will refer to the latter as “the mHealth app” and use the term “the mobile tool” when referring to the tablet with all its apps.

The mHealth app on the tablet computer was developed after an initiative by professionals (2 doctors and 2 nurses) at one of the participating academic PCCs. They saw a need for easy access to information, including patient data, and to carry out all parts of their home care work, including documentation, while in the patient’s home. The HCP took an active part in an initial need and requirement analysis, and the mHealth app was then developed from their ideas. The first author was involved in some of the app’s needs and requirement analysis. Professionals provided feedback on functionality and design but were not further involved in the mHealth app’s development. No structured usability assessments were performed.

The project was carried out through a collaboration between academic PCCs and the IT unit of Region Stockholm, along with the private company Chorus, which was responsible for the technical development of the app. As an innovation project, each participating entity covered its own costs. The aim of the pilot project was to evaluate the tool’s effectiveness and its potential for broader implementation within the region, with the long-term goal of generating both clinical benefits and economic value. The mobile tool was tested at selected academic PCCs, and the responsibility for approval and rollout was shared between the involved academic PCCs and the IT unit of Region Stockholm.

The mHealth app provided mobile access to the EHR used in regional home care. This included, for example, access to patient record entries, information about the patients to be visited during the day, and the possibility to write patient record entries while in the patient’s home, a task otherwise done after returning to the office, as previously described in our study outlining the matching of the functionality of the app to the home care work processes [24]. While the previously published study described the work process in primary care in detail, matching it to the functionality of the study, it did not go into details on the users’ experiences and reasons for the failed implementation. Due to technical difficulties, not all parts of the EHR were accessible through the mHealth app. The app did not provide information on patients’ medications, referrals, or radiology results because the EHR from which the mHealth app collected the medical data did not support this output. However, it was possible to access information on the patients’ previous outpatient and inpatient care visits, diagnoses, and laboratory tests. Thereby, some, but not all the HCPs’ needs were met by the mHealth app. The mHealth app was intended to be further developed, and the implementation was therefore initiated.

The tablet computer also provided access to several clinical decision support systems and a mobile web-based prescription system that is used for some of the patients in home care, thus enabling the HCP to manage part of those patients’ medications. Thereby, the diverse functions of the tablet computer provided HCPs with a multifaceted mobile tool for home care visits.

There were no costs involved for the users. Use of the mobile tool was strongly encouraged but voluntary.

The mobile tool was first implemented in the home care section of the PCC, which was part of the development of the mHealth app and in one of the other PCCs. In both participating PCCs where the mobile tool was first implemented, some HCPs were avid advocates for the mHealth app.

At a later stage, the mobile tool was implemented in the MHCO and the remaining PCCs. There were no advocates for the mHealth app at either of these workplaces. The mHealth app was further developed later to improve stability and implemented and used in in-hospital care for bedside record access in the Stockholm region. In 2020, Region Stockholm ceased all support of the mHealth app, and it is currently not in commercial operation.

Data for this study were collected through focus groups and semistructured interviews with nurses and doctors working in home care where the mobile tool was introduced. The first author (LJH, female) was responsible for the data collection, with the assistance of 2 Master’s students (both female, health informatics students). All 3 had both training in and experience of qualitative data collection, and were supported by the last author (MH, female) with extensive experience in qualitative research. LJH was working as a general practitioner in the region at the time of the study and was well-known to all study participants. An overview of the timeline and data collection is presented in Table 1. Before implementation, a workshop was held where questions were asked regarding the time spent on administrative work related to home visits, experienced stress, use of decision support tools, and the experienced quality of documentation and communication between HCPs.

The mobile tool was introduced and demonstrated during a workshop where participants had the opportunity to practice using it. The demonstration was conducted by guiding participants through the different functionalities step by step, ensuring a comprehensive understanding of its features and applications. This hands-on approach facilitated engagement and allowed attendees to gain practical experience with the tool.

The original plan was to conduct data collection through (1) preimplementation focus groups, (2) individual interviews after 6 months of use, and (3) a second round of focus group interviews after 12 months of use. The different qualitative methods were chosen to gather a rich dataset; the focus group interviews enabled discussion and exchange of opinions between participants, whereas the interviews allowed more detailed exploration of individual HCPs’ experiences of using the mobile tool. We also recognized that individual interviews provide an opportunity for participants to express thoughts and concerns they might be reluctant to share in a focus group setting with other participants present.

Interview guides were developed for individual and focus group interviews (Multimedia Appendices 2 and 3). All interviews and focus groups were recorded and transcribed. Notes were also taken during focus groups and included in the material. All interviews were held in Swedish. The individual interviews lasted 20‐30 minutes and the focus group interviews lasted 1‐1.5 hours. Data collection was conducted where most convenient for the participants, that is, at the PCCs or the MHCO’s office. Data collection was conducted during working hours. Participant checking of the transcribed interviews was not done to reduce the burden on the study participants.

Preimplementation focus group interviews were held in all the PCCs and the MHCO, where the positive and negative expectations were discussed and reflected upon by HCP who would use the mobile tool.

Individual interviews with nurses and physicians were conducted according to plan to capture the staff’s experiences using the mobile tool and its impact on work processes.

Interviews were conducted when the mobile tool had been used for at least 6 months. All interviews were conducted by the first author and the Master’s students in the research team. The Master’s students conducted interviews in the MHCO. One of the Master’s students continued working in the project as a research assistant after graduation and also contributed to the data analysis.

All eligible nurses and physicians working in the setting were invited and accepted to participate in the study. Study participants were approached via email and in-person by the first author through the managers of the MHCO and PCCs. The same participants at each site participated in both the focus groups and interviews.

A second set of focus group interviews was planned at the end of the study, when the mobile tool had been used for 1 year. The PCC involved in the mHealth app’s development started testing earlier than the other clinics, and a focus group interview was held a year after they started using the mobile tool to capture their long-term experiences. When the 6-month interviews were conducted at the other clinics, they had stopped using the mobile tool, and the decision was made not to conduct the planned focus group interviews, as they would not gain further experiences of using the mobile tool.

All interviews were transcribed verbatim and analyzed using a deductive content analysis approach [25]. The CFIR [23] was used to structure the analysis. The data analysis performed in this study does not include any quantification of responses regarding specific categories.

CFIR identifies 5 domains that are essential to implement an intervention [23]:

In CFIR, a domain represents a broad category of factors or determinants that affect the implementation process. A construct, on the other hand, refers to a specific factor within a domain that provides more detailed insight into the aspects of implementation. An overview of the CFIR domains and their constructs is presented in Figure 2.

The 3 authors all actively participated in the analysis. Initial coding was performed by LJH, MH, and one of the Master’s students who continued working on the project as a research assistant after graduating. First, all materials were read and reread, and meaning-bearing units were identified. Each coder coded and categorized the meaning-bearing units individually into 1 of the 5 domains of the CFIR framework. HH joined the analysis at this stage, and the 3 authors discussed the coding until a consensus was reached on the interpretation. Where appropriate, relevant CFIR constructs were used to signify our findings, but we did not systematically search for constructs in the material.

Due to the relatively small number of participants, achieving full data saturation was not possible. However, we ensured sufficient information power by including all eligible HCPs across multiple roles and settings. Thematic consistency across discussions suggests key insights were captured despite the limited number of study participants.

The Research Ethics Committee at Karolinska Institutet has granted ethical approval for this study (DNR 2015/1457-31/5). All study participants provided written or oral informed consent before participating in the study. In accordance with the Swedish Ethical Review Authorities’ requirements, participants were informed that their participation was voluntary and that they could choose to revoke their participation at any time. Transcribed interviews and focus groups were deidentified to protect the participants’ privacy, and confidentiality was preserved when reporting results. Study participants did not receive any compensation for their time.

The 18 participants in the study were all nurses and physicians working in home care in either the home care section of the participating PCCs or the MHCO. A total of 8 nurses and 1 physician from the PCCs (PCC1=2 nurses and 1 physician, PCC2=1 nurse, PCC3=3 nurses, and PCC4=2 nurses), and 8 nurses and 1 physician from the MHCO participated in the study (Figure 1). Participants were predominantly female (17/18, 94.5%), and age ranged between 43 and 64 years. Due to the small number of participants in the study, we chose not to collect and report further detailed characteristics of the individual study participants to ensure their anonymity.

Using the 5 domains of CFIR, we present our analysis of the HCPs’ expectations before implementation and their experiences during the implementation of the mobile tool. We identify the key barriers causing the implementation to fail and the aspects that had a positive influence.

Before the implementation, HCPs’ expectations were high, yet the interviews after 3 and 6 months of using the app revealed that many of the expectations were unmet. Textbox 1 presents an overview of the constructs related to the characteristics of the intervention, with further details from the qualitative analysis presented below.

In this domain, the expectations of the intervention coincided with the actual experiences of the intervention to a large extent. Textbox 2 shows the categories identified in the outer setting domain.

The HCP’s expectations regarding the outer setting were fulfilled to a high degree.

The concerns were mostly addressed, but issues related to the support within the organization were also experienced.

As previously described, the user’s tolerance for challenges differed. Some continued using the mobile tool despite difficulties, whereas others stopped using it. The categories related to this domain are presented in Textbox 4.

Although the participants’ characteristics did not necessarily change during the implementation, their self-efficacy and beliefs about the intervention appear to have been affected.

The implementation process was not addressed in interviews with HCP, as this was not the focus of the original project. We will reflect on the implementation process in the discussion.

The implementation of the mHealth app was unsuccessful, despite reports of positive experiences and potential benefits. The main reason for the failed implementation was that the app did not sufficiently support the users’ needs. Immature technology and unstable technical infrastructure also contributed. The findings from this implementation study highlight the critical importance of aligning digital health interventions with user needs and existing infrastructure, particularly in the context of aging populations and resource-constrained health care systems. For digitalization to be effective and sustainable, strategic approaches must prioritize usability, cross-organizational compatibility, and user involvement throughout development and implementation. Moreover, adequate resourcing—including time, training, and leadership engagement—is essential to prevent change fatigue and to ensure long-term adoption and impact of mHealth technologies.

Before implementation, HCPs had high expectations that the app would facilitate work processes and reduce cognitive workload by allowing real-time access to information. There were also concerns, mainly due to prior experiences of poor internet connectivity and immature and unstable technology, but also that the app might increase the workload, shift time, and focus from the patient, and that the project would amount to nothing. Ultimately, both expectations and concerns were confirmed. Another preimplementation concern expressed by some HCPs was that they would become too dependent on the new technology. This was not brought up again postimplementation, and we interpret this as being due to the failed implementation. The concern could have negatively influenced the HCP’s willingness to adopt the technology, but likely this was not the most important factor for the failed implementation.

Some users, particularly those involved in the mHealth app’s development, found value in the app despite its shortcomings and kept trying time after time, undeterred by the immature technology. However, users who worked in the MHCO and had not been involved in the design process gave up at the first hurdle and did not use the mobile tool. Data on usage of the mHealth app, for example, the number of log-ins per day or number of active users, were not available during the implementation process. Access to such data could potentially have guided the implementation process so that action could have been taken when usage was low [26].

End user involvement in the design process, often referred to as participatory design [27-29], has proven to be essential in both ensuring that the software meets the end user’s needs and supports their work processes, and increases acceptance among the users [30,31]. In addition to the difference in participation during the development process, the MHCO also used another EHR system in their daily work that was not integrated with the mobile system. The benefit of being able to at least begin documentation in the patient’s home was lost to them, further lowering their incentives to engage with the system. Lack of interoperability and integration between eHealth systems has previously been identified as an important barrier to successful implementation [16]. Physicians who made few home visits and used dictation for EHR documentation in everyday work found the mHealth app less useful. Again, the importance of capturing the perspectives of all potential end users in the design process was highlighted [31,32].

There are several lessons to be learned from this. First, that outcome depends on context, and second, that user engagement in both the development and the implementation processes is crucial for a successful implementation. In our study, both lack of user involvement [30,33] and poor interoperability [7,17] hurt the acceptability of the system, contributing to a system that did not support the user’s work processes [24]. The importance of an interoperable digital health ecosystem is stressed in the World Health Organization digital health strategy, defining it as a “digital interoperable information technology infrastructure that is primarily used by the health care community across all care settings, in particular by health care providers, health service providers and patients” [4]. With such an infrastructure in place, the integration and implementation of the mHealth app would likely have been more successful. The lengthy development and implementation process also affected user engagement negatively. Studies of user involvement’s relationship with user satisfaction have indicated that user involvement can overcome frustration from delays and prolonged projects [30]. In our study, we did see higher user satisfaction and tolerance among users who had been more actively involved in the design process, but it was not enough to overcome the implementation challenges completely. Additionally, for less engaged staff, the barrier to start using a new technology was high, and they also had a greater need for additional support, whereas users with high engagement, often referred to as “champions” in the implementation science literature, kept trying regardless and often found new areas of use for the technology. The most common reason for discarding an eHealth service is issues with the technology [19,34]. eHealth services often are premature and thus do not adequately support the work in home care [19]. Technical issues [19] include everything from a lack of suitable infrastructure to usability issues, the latter referring to whether the product provides the right utility and the degree to which it helps users reach their goals in an effective, efficient, and safe manner, while still being easy to learn and remember how to use [35]. Usability of the mHealth app should have been evaluated more extensively before the implementation, for example, through user-driven scenario-based usability testing [36], or expert- or user-driven heuristic evaluations [37,38].

In our case, considerable problems with internet connectivity, immature technology, and a lack of functionality contributed to the low app usage.

Other studies and reviews confirm the challenges described by the users in this study. User involvement, interoperability, reliability of connection, and technology and infrastructure can be considered facilitators if handled appropriately in the implementation of eHealth, or barriers if they are not considered [12,17]. In a review of 43 studies of mHealth implementation experiences from primary health care services, the main results were that health workers appreciated mHealth when it improved feedback, speed, and workflow [17]. Challenges stated were that mHealth sometimes created more work, resulting in some preferring paper instead. Some health workers found the decision support opportunities useful, and others thought it threatened their clinical skills [18]. The aspect of workflow was found substantial for the success or failure of eHealth interventions particularly if workload increased, if workflow was interrupted, and if there was an alignment with clinical processes [17].

CFIR is developed to study implementation processes, not specifically new technology. We chose to include “technical infrastructure” as a new construct in the domain “Outer setting” since the effect of new technology depends on its functionality. A more specific focus on the usability of the eHealth or mHealth intervention could also be included under intervention characteristics, but we chose to use the “design quality” construct here.

The implementation process is crucial for the success of an implementation and for a successful implementation, all stages of the implementation must be properly executed. Our study was not originally designed to study the implementation process, and subsequently, material on some aspects of the implementation process is missing since no questions regarding those aspects were asked in the interviews. We see, however, that the implementation of the mobile tool most likely would have benefited from better planning, execution, support functions, and resources. When implementing complex interventions, feasibility and pilot studies are essential [35,39]. This is especially important when implementing mHealth where lack of usability is often the cause of failed implementations. This could be further specified in the CFIR domain “process” when studying new technology. A specific challenge of implementing the mHealth app in a rural home care area was the physical distance between units and the challenges of keeping day-to-day contact with and providing support to the HCP. The need for a specific facilitator role, that is, a person dedicated to supporting the implementation process and ensuring problems are identified and addressed, was apparent, as is also stressed in other implementation theories, for example, the Promoting Action on Research Implementation in Health Services (PARIHS) framework [40]. In line with the integrated-Promoting Action on Research Implementation in Health Services (i-PARIHS) framework and other implementation models, the facilitator role typically requires a combination of technical, organizational, and interpersonal competencies. Core competencies include basic technical troubleshooting and familiarity with the intervention, skills in user training and coaching, and the ability to communicate effectively across different organizational levels, including leadership. Responsibilities often span the entire implementation process: supporting planning and readiness activities before rollout, providing hands-on assistance and problem-solving during implementation, and ensuring ongoing support and feedback loops during maintenance. While the exact scope and emphasis of these tasks must be adapted to the local context and resources, these elements are commonly highlighted as critical for successful implementation.

There is a strong belief in the potential for eHealth and the introduction of new technology to revolutionize health care and that it will be necessary for health care and primary care in the future [4]. The introduction of new technology, however, often fails [17,41], and there is a risk that premature implementation of new technology will exhaust HCP for future projects and implementation. The challenges and concerns must be considered to move forward successfully [15]. According to the CFIR [23,42] and i-PARIHS [40] frameworks, the individuals’ previous experiences and beliefs about the intervention will impact the implementation’s success. When HCPs are exposed to immature digital solutions with poor usability, we may risk a negative impact on future implementation and innovation projects. To avoid this, we conclude that it is important to ensure that new technology has the necessary functionality and stability before implementation and that there are routines in place in the organization to guide decision-making and implementation processes. Structured contingency planning for technical failures, such as providing clear fallback procedures, timely support, and communication strategies, could sustain user engagement by reducing frustration, maintaining workflow continuity, and reinforcing trust in the implementation process despite technological setbacks.

The study was guided by the CFIR framework, which is a well-established model for understanding implementation processes [23]. This adds theoretical rigor and structure to the analysis, ensuring that a broad range of influential factors is considered. In our analysis, we used the original CFIR framework which was available when we began our analysis. More recent versions of the framework [42] have added new constructs; however, the domains remain the same, and we believe that using the most recent version of CFIR would not have substantially influenced our results.

The use of interviews (both individual and focus group) allowed for an in-depth exploration of the HCPs’ experiences and perspectives. This is particularly useful given the small number of users of the mHealth app, making qualitative methods appropriate for capturing rich, detailed insights. The study also collected data from both nurses and physicians across different settings (PCCs and MHCO), which ensured a variety of perspectives were included in the analysis. The small sample size (8 nurses and 1 physician from each of the two groups) may limit the generalizability of the findings to other settings or health care systems. Additionally, the majority of participants were female, which may impact the diversity of perspectives, yet it reflects the current workforce in Swedish home care. To maintain anonymity, the study does not report detailed demographic information about participants, such as years of experience, which could be useful for understanding how these factors might influence perceptions of the mHealth app. We also do not include any quantification of responses, which means that while rich qualitative data were captured, the relative importance of different factors or themes was not assessed. This could limit the ability to weigh certain barriers or facilitators more heavily than others; however, considering the qualitative nature of our dataset, we argue that this is the best approach as quantifying the results may be rather misleading. Using a validated instrument to measure health care professionals’ self-efficacy could be a way to further understand how self-efficacy impacts adoption and abandonment, and how it may change over time during an implementation project. This should be explored further in future research.

A strength of the study is that the data collection spanned several years, and follow-up interviews after 6 months of using the mobile tool captured longer-term experiences and effects, offering insights into sustained usage and challenges. However, the study took place from 2015 to 2017, with final data collection in 2018, which could render the results of the study out of date. Yet the findings remain highly relevant in today’s digital health landscape. Despite technological advancements, the fundamental implementation challenges we identified, such as poor alignment with clinical workflows, limited user engagement, inadequate training and support, and insufficient organizational readiness, persist in contemporary mHealth and eHealth initiatives. Implementation projects continue to experience implementation challenges across diverse settings, also in large-scale initiatives. For example, recent EHR rollouts in Sweden [43], Norway [44], and Finland and Denmark [45] encountered major disruptions due to user dissatisfaction, lack of preparation, and technological shortcomings. These cases underscore the enduring importance of addressing the sociotechnical and organizational dimensions of digital health implementations.

The first author was involved in the needs and requirement analysis for the mHealth app and also conducted the interviews. This dual role might introduce potential bias into the data collection and analysis process, as prior involvement may influence how the data are interpreted or what aspects are emphasized. Throughout the analysis process, we have been vigilant about this, and the involvement of Master’s students with no prior connection to the project facilitated an unbiased data collection. The first author’s familiarity with the context and the app has also been an enabling factor, to both data collection and by facilitating insights into both the design and implementation processes that would otherwise have been challenging to obtain in retrospect.

These strengths and limitations provide important context for interpreting the findings of the study and considering its applicability to broader implementation efforts.

We conclude that new technology must be stable and have the desired functionality before implementation for success. The functionality of a new technology must support the users’ needs and be sufficiently integrated with other IT systems in the health care organization. How the implementation process is executed is important, but the implementation will not be successful if the new technology does not properly support the users’ needs. In the case presented here, implementation failed because the technology was unstable and thus unreliable, not sufficiently supportive of the users’ needs, and not adapted to the users’ workflow. There were, however, positive experiences too, and mHealth still has a strong potential to support HCPs working in home care. To achieve this potential, we must ensure that sufficient resources are allocated to both design and development, evaluation and feasibility studies, and support and engagement during the implementation of mHealth in the future.