The rising costs of health care services and the increased prevalence of chronic conditions necessitate transformative changes in health care delivery. Deliberations on these changes have focused on access to information and the interconnectedness of health information systems [1]. Providing patients secure access to their health information is a provincial [2-5] and national goal in Canada [6,7], and it is considered critical to the delivery of patient- and family-centered care [2-4]. Surprisingly, many portals have underperformed expectations in Canada [8] and in other jurisdictions [9,10]. Although patients today have greater access to their medical information than ever before, there is a need to identify the best way to provide this information and to understand the impact of the provision of medical information to patients on the health care providers.

To address the issues mentioned above, the Alberta Health Services (AHS), a province-wide health delivery organization, piloted a patient portal called eCLINICIAN MyChart (an AHS branding of EpicCare Ambulatory from Epic Systems, 2014 version) between 2016 and 2019 [11]. The goals of the study were to determine whether MyChart would help patients and their families participate actively in the maintenance and monitoring of their health information and to provide information that could be used to scale up the use of MyChart by documenting factors that both hindered and promoted MyChart uptake.

We aimed to accomplish this goal by collecting data from both users and nonusers at the pilot sites. We focused on factors that promoted the use of MyChart along with the related benefits and barriers to its use, with the intention that this information could be incorporated into the plan for its province-wide implementation. To the best of our knowledge, this is the first study that has considered the perspectives not only of patients and health care providers but also of clinic managers and nonmedical staff, such as receptionists, on the impact of patient portal uptake. The timing of data collection began before the introduction of the portal to identify expectations held by health care providers’ practitioners and managers. This approach to data collection was used to identify previously unknown or unanticipated factors that could limit the potential of patient portals for health care systems, including previously obscure power dynamics among clinical leadership. Portals are likely to have systemic groups of nonusers for reasons that have not been previously articulated. In this study, we sought to clarify why portals can, at times, fail to achieve their transformative potential. In parallel with this work, a quantitative survey was conducted on patient views of the portal [12], and the views of the clinicians and managers presented here complement those findings.

The research question was as follows: What factors impacted the differences in uptake of the patient portal MyChart and what elements were deemed crucial for uptake and access for all end users?

This question was developed by incorporating the principles of diffusion of innovation by Rogers [13] and the theory of technology acceptance by Davis [14]. Roger’s theory focuses on explaining how and why new ideas and technologies spread, whereas Davis’ s theory focuses on the perceived usefulness of a technology to enhance performance [13,14]. By incorporating these 2 frameworks, we were able to consider the various theoretical approaches relevant when considering the uptake of a patient portal within a publicly funded health care system.

This study received ethics approval from the Research Ethics Board at the University of Alberta (Study ID: MS6_Pro00084135) and operational approval for the clinics included in this study from AHS (OA43157, OA43158, OA43159, and OA43160).

Using a qualitative comparative case study design [15-17], we performed 56 in-depth semistructured interviews with clinic managers, health care providers, nonmedical providers (receptionists), and patients from 5 clinics that participated in the pilot. Comparative case studies involve the analysis and synthesis of similarities, differences, and patterns across 2 or more cases that share a common focus or goal [15]. In this study, each of the 5 clinics constituted a case. A comparison of cases provided an opportunity to learn more about factors that played a role in the uptake during a real-time patient portal implementation from the standpoint of study participants rather than from existing frameworks.

Alberta, Canada, has a publicly funded and managed health care system based on the principles of universal access to medically necessary services. AHS [18] is the largest provider of health care services in Alberta, delivering care to more than 4 million people. In 2016, AHS introduced a patient portal called MyChart that was connected to the central electronic medical record system called eCLINICIAN. This was carried out through a pilot where several self-selected clinics decided to participate; however, individual health care providers within the clinics were free to participate or not. MyChart allowed patients to view appointments, medical test results, and medication therapies and to communicate with their health care providers through a computer or an app on a phone or tablet. It was developed by the Epic Systems Corporation (Epic) and customized to fit the needs of AHS. The pilot was led by the eCLINICIAN MyChart Working Group, comprising physicians from both family medicine and various specialties, technology experts, policy makers, and health care administrators.

Following ethics approval, 5 clinics in the pilot were selected for this study based on the number of patients enrolled in MyChart per clinic and the length of time each clinic had been using it. These clinics were considered minimal and safe users of the nonportal functionalities of the Epic product (eg, eCLINICIAN). Three of the clinics were specialist clinics, 1 clinic was a community-based family practice clinic, and 1 clinic was a family practice clinic hosted within an AHS facility.The AHS family practice clinic was selected because although in principle they agreed to participate, they never enrolled any patients. The key features of the cases are outlined in Table 1.

Before deciding on the required study sample, the project team had discussions with the MyChart project team and the staff of several clinics in the pilot to understand how its use evolved in each setting. On the basis of those conversations, MA developed a participant recruitment framework (Figure 1), which was reviewed by KO, TS, DM, and MB and then finalized.

On the basis of this framework, we decided to recruit clinic managers and health care providers who did or did not participate in the pilot and patients who used or did not use MyChart, including proxy users. MyChart permits proxy access to family members who provide care to their loved ones.

Clinic managers and health care providers were recruited using a purposive sampling [19] approach. TG contacted each identified clinic, requested an introductory meeting with the clinic manager, and responded to any follow-up concerns. The clinic managers invited health care providers (specialists, general practitioners, and/or nurses) to participate in these introductory meetings. MA developed the presentation and recruitment materials. DM and TS reviewed the materials. MA and MB attended these meetings, described the study, and discussed recruitment strategies for health care providers and patients within each clinic. After answering any questions, MA obtained written consent from health care providers and other clinic staff who were willing to participate in the study. MA set up interviews with clinic managers and health care providers. During the interviews, we identified that only 1 clinic implemented the MyChart web-based booking and canceling appointments function, so all 4 receptionists (nonmedical providers) in that clinic were interviewed.

Patients were recruited using a recruitment card (Multimedia Appendix 1), which was handed out by health care providers at each clinic over a 2-week period. They had provided the card to all patients they saw during those 2 weeks. Patients contacted MA directly if they were interested in participating.

MA scheduled interviews and explained the study, answered questions, and obtained written consent from those who agreed to participate. The participants then completed a demographic information form and were assigned a deidentifying code used to maintain anonymity and track data. All recruited participants took part in an individual semistructured interview (conducted by MA face-to-face or by telephone) lasting between 30 and 45 min. Notes and reflections were written by MA immediately after each interview. The semistructured interview guides (Multimedia Appendix 2) were developed by MA, based on the research question, and then were reviewed by the other authors and revised based on feedback. To determine whether the interview guides required adjustments, MA evaluated the recordings of the first 2 interviews per group. No changes were made. All interviews were recorded digitally, transcribed verbatim, checked for accuracy, and then uploaded to NVivo 11 (QSR International) to facilitate analysis. Data collection and analysis occurred concurrently until no new dimensions of the participants’ experiences were identified [20].

The data collected guided the structure of the subsequent analyses, following the principles of inductive thematic analysis [21-23]. MA cleaned all transcripts and then read each transcript several times, coded transcripts for key ideas related to the research question, and grouped codes into preliminary categories. KO reviewed all codes and categories and discussed them with MA. Preliminary categories across cases and participant groups were organized according to recurring and emerging themes by MA and reviewed by KO and TS. As is common in qualitative research, the findings were built from the data [24]. The findings were then compared with those of other studies.

Several processes were undertaken to maintain the trustworthiness of the data [23], including establishing credibility by recruiting a diverse sample with different and sometimes opposing perspectives and maintaining a comprehensive audit trail in NVivo 11. To ensure transferability of the data, detailed descriptions of the sample and thick descriptions of findings were developed. Analytic rigor was enhanced through regular meetings of the research team to discuss emerging findings. As each interview was completed, all available interviews were reviewed collectively. Data saturation was reached after 25 patient and 20 health care provider interviews, but 2 more patients and 1 more provider were interviewed to ensure that important information was not missed. The demographic data were analyzed descriptively and added as case nodes in NVivo to explore factors associated with the adoption of MyChart.

Our plan was to recruit patients who would fit one of the following categories: user (has access to MyChart and has used it at least once in the past 6 months), declined (offered but declined to sign up), did not continue to use (signed up but did not use), and proxy access user. We were able to recruit only 1 patient who declined and none who did not continue to use, as we had no means of obtaining information about discontinued usage.

We also recruited 1 caregiver who used MyChart under the proxy provisions and several caregivers who used the family member accounts. Table 2 shows the patient respondents’ demographics and reported conditions.

We were able to recruit several health care providers who were nonusers (1 provider elected not to participate in the pilot, 1 medical resident who was not aware that MyChart was part of the clinic where he or she was working, and 3 medical office assistants who were familiar with MyChart; however, as a result of the implementation process for MyChart in that particular clinic, they were not involved directly with MyChart). The specialty physicians were endocrinologists (n=3), neurologists (n=5), and gastroenterologists (n=5). Table 3 shows the health care providers’ and clinic managers’ demographic information.

Overall, 56 participants were recruited (Table 4; to see interview codes and the demographic characteristics per participant group, see Multimedia Appendices 3 and 4).

We coded each group of interviewees (clinic managers, health care providers, nonmedical staff, and patients) separately. As we were reviewing the transcripts and coding them, it became evident that the concerns raised by nonusers were also raised by users of MyChart. Therefore, information collected from users and nonusers was aggregated. Textbox 1 shows the identified themes.

The themes from each sample group are described in more detail below, with direct quotations used to support the analysis.

We report the uptake and adoption of an initial phase of a patient portal introduction among 5 health care clinics that were, at the outset of the introduction, judged to be clinics where high uptake would be found. In contrast, 2 clinics showed high uptake, 1 clinic showed moderate uptake, 1 clinic showed low uptake, and the last clinic was categorized as a clinic without use. We identified several factors related to the uptake of the patient portal, including satisfaction and/or dissatisfaction with the current communication tools, performance expectancy, facilitating conditions, behavioral intentions, and use behavior. All these factors had an impact on the level of uptake in each setting and created different opportunities for the end users. Table 5 summarizes our findings about each clinic by considering our findings.

In our study, dissatisfaction with current health care communication tools appeared to be the primary factor that drove patients’ and providers’ decisions to consider the use of MyChart. Once they had made this decision, uptake was dependent on performance expectancy, which included understanding how to use the technology and valuing the additional functionality that it offered. Furthermore, although the themes emerged from the interview data, they aligned with the theory of technology acceptance by Davis [14] and included perceived usefulness, perceived ease of use, barriers to ease of use, and future usage [14]. For example, theme 1 from the patient interviews was my health, my responsibility, but I need the information to do that, which correlated with perceived usefulness [14], as shown in Textbox 2.

The conditions that facilitated understanding of the technology varied across clinics. Some clinics formally designated assistance with MyChart use to staff in their clinic. All clinics had high or medium uptake of MyChart. In other clinics, staff decided not to offer MyChart to some patients, which had an impact on the uptake of MyChart. Reasons are not entirely clear and require further investigation. Some patients who understood how to use the technology and valued it became users, whereas others did not. The main reason for discontinuing use appeared to be the lack of an immediate health care concern [25].

Clinic managers were foundational to the uptake of MyChart (Figure 2), a finding that has not been discussed in the literature to date. They made decisions about whether to present the technology to providers. Our data suggest that this decision was based on an informal assessment of satisfaction with the current communication tools. Clinic managers who declined participation in the pilot indicated that their clinic patients and providers would not be interested in trying the technology. Figure 2 summarizes the impact of the clinic managers on the uptake of MyChart.

The actions of the clinic managers could be observed through Roger’s theory of diffusion, as they served as early adopters by assessing the advantages or disadvantages of the innovation. They determined the perceived efficiencies, its compatibility with the existing system, the complexity, and, ultimately, the benefits and unintended consequences of the innovation [13].

Usage and enrollment depended on how patients were informed about the portal. Patients in this study reported variable understanding of the portal following the introduction by clinic staff. Similar to the study by Kim et al [25], a quick introduction did not appear to be valuable to patients. In this study, participating health care providers reported not having sufficient time to explain the portal to patients. In clinics with higher uptake, a short introduction became incorporated into the clinic visit workflow, which allowed for more sufficient means in enticing patients to sign up.

Knowledge about the technology was another important component that led to increased or decreased usage. Ryan et al [26] found that patients and health care providers had positive views about patient portals but that older patients had less knowledge about patient portals. In this study, older participants reported that they took the time to learn and understand the system once they decided to enroll. We found that patients had to learn the system by trial and error. When patients were unable to navigate the sign up and authentication, the use of MyChart became impossible. When patients accessed medical information in MyChart, they reported becoming more active in the management of their health, which allowed them to see their health trends and identify opportunities for self-care. Patients came prepared with specific health questions because they could spend less time going over less valuable health information (eg, lab results) and making better use of scarce time with their care provider. This finding was similar to the results reported by Dendere et al [27].

Some research teams have reported a relationship between concerns regarding privacy and technology uptake [28-36]. In our study, participants did not report perceiving significant risks to having the confidentiality of their health information violated. Participants reported a sense that the health system would maintain appropriate controls and procedures to protect the security of their health information.

Although not all health care providers saw the value of patients having access to the same information, participants championing use of MyChart experienced more efficient ways to connect and communicate with their patients. This finding is in contrast to the findings reported by others who have studied patient portals [27,37-40] and reported that portals increased workload. For example, Koivunen et al [41] described the negative views of nurses about patient portals. Our findings are consistent with the work of other research teams who reported that nurses saw patient portals as an opportunity to improve communication with patients [42-44].

Furthermore, theories of how patient portals support health care system transformation may not draw upon crucial empirical data. The current findings emphasize that without a shared understanding of the purpose of patient portals and some dissatisfaction with existing communication methods, uptake lags. We posit that these aspects will be crucial as patient portals introduced in real-life care settings often accompany other technologies for clinicians and managers to navigate [44].

We agree with Wiljer et al [45], who described a need for a culture shift, where access to medical records is a fundamental right of every patient. A shift of this magnitude could only be accomplished with broad support from policy makers, health care providers, health administrators, and patients. Without adequate support for patients as complete partners in the management of their health, patients would be left to take actions based on incomplete and possibly inaccurate information from untrustworthy sources. Clinically useful uptake of patient portals is an essential enabler for patients as partners in managing their health and transforming the wider health system.

There were several limitations to this study. Patient interviews were only conducted with patients who volunteered to participate and were attending an appointment with a health care provider who was also participating in the study. The information provided by participants may not have been recalled accurately by them. In addition, although all efforts were made, we were unable to identify and recruit participants who did not continue to use MyChart as we had no means to obtain information about discontinued usage. Finally, the findings should be generalized with caution to other health care systems and to settings that use other types of patient portals.

This work exposes a number of factors that have an impact on the uptake of MyChart. We found that dissatisfaction with nonportal-based communications was a foundational element that likely needs to be addressed before more advanced goals can be reached. We also found that clinical managers were key gatekeepers in the uptake process. Once a clinic manager decided that the clinic should use the portal, the designation of a staff member to help patients become more familiar with the portal appeared to increase uptake.