In this qualitative study, we asked participants from various backgrounds how they encouraged users to adopt a patient portal and what they had learned from their efforts, in order to understand what stimulated or hindered the adoption process. We conducted 22 multi-actor, semistructured group interviews and held four structured focus groups to check, enlarge, and enrich our findings [24-26]. Table 2 lists the different forms of data collection.

All interviews, both group and individual, and focus group sessions took place in a hospital or online. All the hospitals included in this study are participating in a national program—VIPP (Versnellingsprogramma Informatie- Uitwisseling Patiënt en Professional) [27]—initiated by the Dutch Hospital Association and the Dutch government. VIPP is the Dutch government’s financial incentive program to support information exchange between patients and professionals through patient portals. The aim of the VIPP program is to give patients online access to their medical data, either through a patient portal or a PHR [27]. Information technology (IT) suppliers with a commercial interest deliver patient portals and PHRs. Given that Dutch hospitals are free to choose any supplier for their patient portals or PHRs, the hospitals in this study use different patient portals. The portals might differ in their available functionalities, but they all offer patients online access to their personal health information [6]. Additionally, hospitals had different aims for their patient portals; only the VIPP program aims were similar for all Dutch hospitals. Hospitals receive financial support based on their achievement of specific national VIPP aims; for example, “In the past 30 days, 25% of all patients (based on DRG [diagnosis-related group] contacts) logged in to the patient portal or the link to a PHR” [28]. How hospitals achieve these aims is left up to the hospital.

There are three categories of Dutch hospital: academic, teaching, and general. Academic hospitals were excluded from our study because at the time they were not participating in the VIPP program. We first determined inclusion criteria for general and teaching hospitals, aiming for a diverse study group. Based on user statistics of patient portals (low and high), geographical differences (rural areas and cities), variation in patient portal suppliers, and usage of the patient portal (more or less than one year), the researchers (PH, AMWJ, and BP) made a selection of targeted hospitals.

In total, 15 of 64 Dutch hospitals (23%) were approached, of which 12 (80%: 10 teaching and 2 general) agreed to participate. Reasons for not participating included “already participating in another study” (n=1) and “too busy with implementation and fear of not meeting VIPP deadlines,” which would mean losing financial support (n=2).

At each hospital or via the Zoom online platform [29], the researchers arranged individual and group interviews in close collaboration with the person running the implementation of the patient portal at that particular hospital. Most often, this person was the project leader who selected participants according to a predefined list of three roles:

In total, 69 participants were included (see Table 2) if they were older than 18 years and had experience with developing and/or using a patient portal. The researchers emailed invitations to participants of the individual and group interviews; groups ranged from 2 to 6 participants. In 10 of the participating hospitals, we organized one or two group interviews; in the remaining two hospitals we conducted one individual interview for logistical reasons.

The aim of the focus groups was to check, enlarge, and enrich our results. Two of the four focus groups were held with project leaders and project staff of hospitals. They joined one of two self-selected focus groups organized during an educational meeting of the VIPP program (n=14 and n=28, respectively). For the other two, we used targeted sampling, selecting experts from patient portal suppliers (n=5) for the third group. The fourth group included macro stakeholders: Ministry of Health, NICTIZ (Nationaal ICT Instituut in de Zorg), the center of expertise for eHealth, health insurance companies, and scientific experts (n=6). All focus group members were invited to take part by email.

Qualitative data were collected on-site in the hospital (n=59) or online via the Zoom platform [29] (n=10) in the fall of 2018. The Zoom platform enabled the inclusion of hard-to-reach, geographically dispersed participants [30]. Group interviews lasted an average of 72 minutes (range 53-88) and the individual interviews lasted an average of 53 minutes (range 44-65). One researcher (PH) conducted all the interviews, following a predefined topic list (see Multimedia Appendix 1) [31] that was based on a search of the literature and discussed among the research team.

The four focus groups lasted an average of 82 minutes (range 71-88) and were steered by a Microsoft PowerPoint presentation explaining the findings of our study. No revisions were made in the presentation between focus group sessions, ensuring that varying opinions were heard before conclusions were drawn [32]. Each focus group was run by two researchers (PH and BP or PH and AMWJ) complementing each other: one moderating and the other taking notes.

During both group interviews and focus group sessions we encouraged the exchange of heterogeneous views that provided insights into similarities and differences in the opinions and experiences of the various stakeholders [26,30,31]. We also invited the participants to challenge each other’s views [33], to explore the implications of their thinking, and to articulate their sometimes-implicit assumptions about the adoption process. This method generated new insights through group interactions.

We audio-recorded the on-site interviews and focus groups and video-recorded the online sessions. All interviews were transcribed verbatim. Analysis, comprising six phases [34], was not linear but a recursive process. First, each individual researcher gave the transcripts a close reading. Second, one researcher (PH) developed codes for the interesting parts of the data. Next, three researchers (PH, AMWJ, and BP) independently developed themes, reaching consensus on a list of relevant themes (eg, communication channel, ambassadors, and patient participation [35]) through discussion. Fourth, one researcher (PH) read the transcripts again. Fifth, using the list of themes, one researcher (PH) performed thematic coding, which two other researchers (AMWJ and BP) checked. In the final phase we (PH, AMWJ, and BP) started writing the results [34].

The Medical Ethics Review Committee of Erasmus Medical Center (Erasmus MC) approved our research proposal (MET-2018-1531) and checked if we were General Data Protection Regulation (GDPR) compliant. All participants were asked to sign an informed consent form.

Our study aims to provide insights into the adoption process of a patient portal by patients and HCPs. The efforts of hospitals to stimulate the use of patient portals can be categorized under three themes: (1) informing patients and professionals about the patient portal, (2) embedding the patient portal in the daily routine of HCPs, and (3) adjusting the portal to patients’ needs in the initial and continuous development process. Below we present our empirical findings for each of these themes.

This qualitative study focuses on patient portal adoption processes by patients and HCPs in a Dutch hospital context. Overall, our results show that the adoption of patient portals is more dynamic than presented in theoretical models and the literature. Greenhalgh et al’s linear adoption stages (ie, preadoption, early use, and established users) [16] seem useful in studying adoption by individuals, but hospital patients and HCPs are in different adoption stages. Consequently, an organization cannot simply move through sequential stages; it needs ongoing effort to be put into informing, embedding, and adjusting to patients’ needs. Their focus on individuals rather than the organizational context is also a criticism levied at the theoretical models (see Table 1) [14,16,17].

All participating hospitals seem to be experimenting with stimulating adoption of the patient portal. They are trying to create effective communication strategies, looking for the best way to embed the portal in daily routines and adjust to its patients’ needs. As yet, they have not found the best way of encouraging portal use by patients and HCPs. Here we explain the implications of our results.

Our study shows that hospitals are experimenting with many communication channels (N=23), mostly ones that are already in use. Despite efforts by communication departments, it seems that portal adoption is still quite a challenge. It seems that hospitals do not know which channels are most effective for which target audiences and what the right timing is for their communication campaigns. Looking at their communication strategies, we found that hospitals choose to emphasize the benefits of portal use and hardly mention the potential disadvantages or risks. According to Greenhalgh et al [16], intended adopters must know the consequences of adopting a patient portal to become established users. If intended adopters are not informed of the potential disadvantages, then they cannot oversee all the consequences of using the portal, for example, the risks. However, the financial incentives of the VIPP program may explain the positive promotion strategy. If hospitals do not attain a certain adoption percentage (ie, 10% or 25%) they will have to repay their VIPP grant.

Our results show that hospitals invest in HCP adoption through peer-to-peer influence. However, focusing on the enthusiastic HCPs can mean that the less-motivated HCPs will lag behind. That a patient portal often does not reduce the burden of HCPs (ie, it only means extra work) and that it is not embedded properly in work routines can hinder adoption. The specialism-focused approach offers a way of encouraging patient portal adoption by HCPs. This experiment with portal embedding would be interesting to study in other contexts to see where and how it could lead to better embedding of the patient portal. A possible disadvantage of this approach could be that patients will not understand why specialisms are in different adoption stages (ie, patients can make an online appointment with one specialism but not for another).

Another principal finding is that hospitals are struggling to adjust the portal to meet patients’ needs in their continuous development process, although all seem to find this important. According to Greenhalgh et al [16], it is vital that intended users get the opportunity to refine the portal so that they will not drop out in the early adoption stage. Hospitals are using various ways to adjust to the patients’ needs in ongoing portal development, without knowing which one is most effective in which phase and for what purpose. It would be interesting to do more research on how patients can participate in portal development, including efforts to stimulate adoption.

Nonusers of patient portals could be studied further. Previous studies show that nonusers have various reasons for not adopting the portal [36,37], including a preference to speak directly to their HCP, the level of their communication skills [36], and their concern for privacy and information security [36]. Such studies would show that hospitals are taking nonusers’ concerns seriously and, at the same time, could produce insights valuable to exploring whether and how the patient portal could be made useful to them.

A remarkable finding is that portal suppliers do not include patients in their development process. The suppliers see it as their clients’ responsibility to give voice to patients’ wishes, but the focus group discussion did make them rethink this. This means that hospitals must explain to the supplier how they should make the portal more user-friendly for patients. Because of the variation in hospital context and portal suppliers, this could explain the disappointing adoption by patients.

Our study has four limitations. The first is that the adoption processes in the hospitals we studied might be somewhat unusual due to the financial incentives of the national VIPP program. Conducting similar research in other countries would, therefore, be interesting and could also teach us more about the contextual, including cultural, factors that influence hospitals’ efforts to stimulate adoption.

The second limitation is the way we recruited hospital participants. Using our own research networks may have biased our sample. However, our recruiting process resulted in a good variation in the mix of included hospitals.

The third limitation is that we only included teaching and general hospitals, given that academic hospitals follow another implementation program. Also, during the study period they were not participating in the VIPP program and, therefore, could not be compared. However, the inclusion criteria context of the studied hospitals varied greatly to include different kinds of hospitals and patients.

Last, this descriptive study shows the efforts that some Dutch hospitals have made to stimulate adoption of a patient portal. We did not study whether the undertaken efforts led to an actual increase of the adoption of patient portals. A further study on the effectiveness of these efforts is recommended.

In recent years, many theoretical models on the adoption of information technology have been developed [9,10]. These models show which variables are important for the adoption of a technology; for example, perceived ease of use, defined in the Technology Acceptance Model (TAM) as “the degree to which the person believes that using the particular system would be free of effort” [12]. However, these models are not explanatory and do not provide the know-how to stimulate patient portal adoption [9,10]. Consequently, we suggest future research should not focus on models, including new ones, but should deal with actionable knowledge for practice [38]. Action research can be used to study the adoption process and the embedding of patient portals in daily practice [39].

Communication experts support the hospitals’ choice to use a cross-media promotion to inform patients and HCPs about the patient portal [40]. Explaining the benefits of using a patient portal is especially important for promotional messages [41]. However, open dialogue among HCPs and project leaders and staff is also vital because it illuminates the professionals’ perspectives on portal development [41]. Earlier research shows that ignoring doubts while trying to convince others to use a technology may produce negative energy and a reluctance to use the portal [41,42]. Further research is required to find the most effective hospital communication strategies for encouraging patient portal adoption for patients and HCPs.

Our study showed that embedding a patient portal in the daily routine of HCPs requires changing their work processes. Earlier studies suggest that hospitals need to make extra time available to HCPs so that they can change and learn new work processes [41]. Research shows that portal use by patients may increase when HCPs are active on the patient portal and it is embedded in their work processes [43]. Research suggests training can benefit the adoption process [41], yet only some hospitals organize courses that explain how the portal works. As a result, some HCPs lack familiarity with portal functionalities [42].

The literature reports several ways of using feedback to adjust the portal to patients’ needs in the continuous development of patient portals. These studies could help hospitals struggling with this. It is important to include patients at the beginning of ongoing development of patient portals [1,44,45]. Examples include co-design, where patients help identify the project based on personal experiences in collaborating with the clinician [45], and participatory stakeholder co-design, where patients and clinicians are equal stakeholders in the whole project [45]. Vulnerable patient groups, such as disadvantaged older adults, should be given special attention in the process of cocreation and user testing [7]. This is an important issue for future research.

Patient portal adoption processes are not just about implementing the technology. They require human interaction in a multitude of ways. Our study reveals three key findings for the adoption process: (1) informing patients and HCPs about the portal, (2) embedding it in the daily routine of HCPs, and (3) adjusting it to patients’ needs in the continuous development of the portal. Our paper provides rich insights into the complexity of the adoption process and gives examples of efforts to stimulate the adoption of patient portals. Our findings help to translate the relatively abstract factors mentioned in the theoretical models to the everyday pragmatics of eHealth projects in hospitals.