We conducted a systematic review of research that discusses physicians’ views on patient portals in the acute care setting. The Sample, Phenomenon of Interest, Design, Evaluation, Research (SPIDER) framework was used to define the inclusion criteria (Table 1). Studies that combined physicians’ perspectives with other health care professionals, such as nurses or physician assistants, were included if physicians were explicitly included in the methods. No limitations were placed on physician specialty, geographic location, or years of practice.

Qualitative, quantitative, and mixed methods studies were included under the criteria that they addressed physician engagement or perspectives on inpatient portals, were published in the English language from 1994 through the last interim search (November 2021), and included search keywords.

Articles focusing on nonphysician clinicians or staff were not included in this review. Outpatient portal and electronic health records discussions were also excluded from this review. Other reasons for article exclusion included not written in English, not about patient portals, focused solely on open or shared notes without addressing any other components of patient portals, and being a description of a study protocol that did not include any results.

We performed the initial database search on February 18, 2021, on PubMed, MEDLINE, Web of Science, and Scopus. An interim literature search across all 3 databases was conducted on November 29, 2021.

The search strategy included a set of keywords relating to patient portals and physicians’ perspectives of patient portals. Keywords were developed using medical subject headlines and derived from scoping articles related to the subject. Search terms included the following: “physician satisfaction” OR “physician satisfaction” OR “physician utilization” OR “physician perceptions” OR “physician attitudes” OR “physician engagement” OR “physician perspectives” OR “physician barriers” OR “physician factors” AND “Patient Portals” OR “patient portals” OR “patient web portals” OR “patient health records” OR “portal adoption” OR “personal health record” OR “online portals.” The search strategy was adjusted for each database (Multimedia Appendix 1). The protocol for this study was registered with PROSPERO (The International Prospective Register of Systematic Reviews; ID #CRD42021236228).

Gray literature was not included. For opinion articles, editorials, and literature reviews discussing inpatient portals, a hand search of references was performed. This did not yield any additional studies to the primary search.

After database search and duplicate removal, articles were imported to Rayyan [11]. Title and abstract screening was done independently by 2 authors, KLB and CC. Conflicts were resolved through discussion between the 2 reviewers upon completion of the title and abstract screening stage until consensus was reached. Included articles were imported back from Rayyan to EndNote (Clarivate). This process was repeated for the interim literature search.

Full-text screening was then performed by KLB and CC. Disagreements during the study selection process were resolved by discussion to achieve consensus.

Quality of studies was assessed using the JBI SUMARI Critical Appraisal Tool (Joanna Briggs Institute) for the appropriate study type. Checklists for qualitative research and analytical cross-sectional studies were used [12,13]. This bias assessment was completed by 2 reviewers, KLB and CC, with conflicts resolved by consensus.

The JBI SUMARI data abstraction form was used for each paper. Data collected included authors, year, methods for data collection and analysis, country of origin, phenomena of interest, setting, participant characteristics and sample size, description of main results, and reported themes. Extracted data were coded by authors KLB and CC. Following initial extraction, data were exported from JBI SUMARI to Microsoft Excel for consolidation and synthesis.

Themes were identified inductively from qualitative and cross-sectional studies as reported by the original authors if available. Otherwise, themes were identified during data extraction and coding by KLB and CC. Original study themes were then categorized and simplified by discussion until consensus was reached by all authors. Data were synthesized using narrative synthesis and reported according to thematic categories.

Reporting of the results from this systematic review was guided by the PRISMA (Preferred Reporting Items for Systematic Review and Meta-Analysis) statement (Multimedia Appendix 2).

From the initial literature search, a total of 3650 references were retrieved across all 3 databases and imported into EndNote 20. After 352 duplicates were removed, 3306 articles underwent screening based on the title and abstract. The interim literature search retrieved 549 new references, and 83 duplicates were removed.

A total of 226 articles were sought for full-text screening, of which 24 articles were unable to be retrieved. A total of 202 articles underwent full-text screening by KLB and CC. After screening was completed, 193 reports were excluded and 9 articles were included. Details regarding the search and selection process from the initial and second search were combined in a PRISMA flowchart (Figure 1). A κ value of 0.579 from the initial title and abstract screening showed a moderate level of agreement between the 2 reviewers [14].

Critical appraisals based on the JBI SUMARI Critical Appraisal Tool are summarized in Table 2 and Table 3. Most qualitative studies had an overall low risk of bias. Several of the cross-sectional studies did not address confounding factors and had a moderate risk of bias. All 9 articles were considered suitable for inclusion in the final review.

The initial and interim search captured studies that evaluated physicians’ perspectives of inpatient portals pre- (4 studies) and postimplementation (5 studies). According to country, 7 originated from the United States, 1 originated from Argentina, and 1 originated from Saudi Arabia. There were 2 studies from pediatric hospitalizations, and 1 from the intensive care unit setting. The studies collected included research categorized as quantitative (n=5) and qualitative (n=4). Table 4 shows the data extracted from the studies.

Participants in the quantitative studies included 1288 clinical team members. This included 375 physicians, among whom 34 were specified as resident physicians. The remaining included physician assistants (n=17), nurse practitioners (n=3), nurses (n=680), clinical support staff (n=205), and pharmacists (n=8). Quantitative data were collected via surveys. Participants in the qualitative studies included a total of 59 physicians, including resident physicians (n=28) and practicing physicians (n=31), a few of whom were specified as hospitalists (n=6). Qualitative data were collected via interviews and focus groups.

Common themes were addressed by participants in both quantitative and qualitative studies. These included perspectives on communication and privacy, portal functionality and patient use, and workflow.

This review identified various themes that emerged from studies of physicians’ perspectives on inpatient portals that were generally consistent with studies of outpatient portals. Recurrent themes included communication and privacy, portal functionality and patient use, and impact on workflow (Table 4). Consistent with other patient portal studies, physicians who have already experienced patient portal use in practice held more positive views, while those without experience of the portal appeared more hesitant about its implementation.

Both qualitative and cross-sectional studies showed that physicians, like other clinical team members, believed the inpatient portal helped patients access information more readily and could promote patient autonomy as well as patient safety, as some patients identified errors in the medication lists and documentation. Managing patients and families’ expectations about information and communication are important to ensuring patient preferences are respected and privacy is maintained.

Physicians without prior experience noted communication concerns about allowing patients to view their laboratory results or notes because it may cause unnecessary stress for patients [20,21,23]. Care team workers, such as nurses, nurse managers, and unit clerks shared similar sentiment toward providing patient access to such information, noting enhanced communication because of portal usage [24].

Postimplementation, physicians appeared to be less concerned about causing patient anxiety that would negatively impacting their workflow in turn [17,19,21]. This complements studies showing that patients are amenable to receiving laboratory and other information through patient portals [25,26]. Some patients prefer to receive information before discussing with the inpatient care team, so that they have an opportunity to formulate more cogent questions for their physicians [25].

This dichotomy was also seen in workload impacts. Inpatient portals ultimately did not seem to negatively impact physician workflow [18] or increase the workload as feared [17]. Some of the increase in workflow was due to needing to ask for help with technical support for patients. Technology concerns were also noted by studies of other clinical team members [24]. Physicians who felt they did not receive sufficient training on portal use themselves were not as optimistic about incorporating the inpatient portal into their current workflow compared to nurses and clinical staff [16]. As noted by other members of the care team, increased hands-on training would be beneficial, as it would highlight the value of the portal and encourage portal usage [24]. In the outpatient setting, physicians and other health care providers have brought up similar concerns over the lack of training and issues with portal usability as a barrier to portal usage [27].

As a result of the COVID-19 pandemic, medical services have become more accessible online as practices implement telemedicine appointments [28]. With such services becoming more broadly available, comfort with virtual health management and patient portal use is expected to increase. The surge in popularity of tools like patient portals requires consideration of physician and provider perspectives. Inpatient needs are more acute, and the availability of nearly real-time information including medication schedule’s effect on patient participation should be explored further. Benefits of both expected features, like medication reconciliation, and unexpected features, like meal ordering, should be considered in future portal development.

Patient adoption of portals is heavily influenced by physician endorsement. When physicians are concerned about increasing work burden from integrating patient portals, they are less likely to encourage use or discuss these platforms with patients, thus reducing patient enrollment and usage [29]. The findings of this review suggest that most of these concerns came from physicians who had no prior experience with patient portals [19,21,30], while physicians with hands-on experience found the portal had little impact on their workload [18]. This effect merits direct study, as it was not explicitly addressed and has implications for how physicians are educated about anticipated portal workflow.

Addressing preconceived notions about portal usage, by providing better training for instance, may help curb pessimism about these digital tools. Improving physician understanding of patient preferences for receiving information can be helpful. Involving physicians during the portal design and implementation process may also allay some of the concerns about workflow and usability.

As inpatient portal research continues to evolve, further research is needed to address how inpatient portals impact quality of care, existing health disparities, and patient engagement. Evidence has shown lower patient portal use among economically disadvantaged populations due to a lack of technology skills, lack of health literacy, lack of English proficiency, preference for in-person communication with providers, and security concerns [31,32]. One study found that patients with higher educational attainment and higher health literacy were more likely to register for the patient portal. After registration, however, health literacy did not seem to affect frequency of accessing the portal [33]. Acute care episodes may be an opportunity to increase health literacy using inpatient portals through assisted registration processes and education. Research on the continuum of outpatient to inpatient portal use is important to identify other factors that may increase portal use [25,34].

Furthermore, since this review identified conflicting perspectives regarding the significance of educational resources for patients through the portal, more research should be done on the types of educational resources that may be valuable to patients and physicians alike.

It is also important to highlight the accessibility of these portals in both inpatient and outpatient settings, and how privacy concerns carry over to patients who have a caregiver. Patients have reported that while certain features of the portal, such as medication information and appointment scheduling, have helped their caregiver to provide better care for the patient, they would like to control what information is shared [35]. To better cater to this population, further studies are needed to inform development of features, such as creating proxy accounts, and how that may affect communication with clinicians. Training on portals among patients and caregivers can encourage use and dispel concerns about technology and security risks. Allowing patients to set boundaries on what information can be accessed by caregivers will also be important for protecting patient privacy [36].

As systems within the hospital become better integrated and more interoperable, there may be opportunities to provide patients with anticipated testing and procedure information to inform their plan as orders are placed, scheduled, or delayed. This may also improve patient satisfaction related to undefined wait times and perioperative delays.

First, although a comprehensive search was performed, this review is limited by the fact that there is only a small number of studies included. There were nearly the same number of pre- and postimplementation studies. With further development and maturation of inpatient portal systems, future postimplementation studies will be able to provide broader practical understanding of issues related to workload and communication.

Second, this study also did not ask questions specific to open notes. Open note–specific studies, such as a survey conducted by Ralston et al [37], demonstrate that clinicians’ attitudes toward open notes changed drastically pre- and postimplementation, as the percentage of physicians perceiving open notes to be beneficial changed from 29% to 71% after implementation [30,37,38]. Studies have also reported that allowing access to notes and clinical data in practice alleviated stress among parents with children who are hospitalized [30,39], and thus physician fears of causing confusion were unrealized [40]. In the outpatient setting, it was found that primary care physicians who shared the same concern about patient access to notes and results ultimately believed that the benefits outweighed their fears, citing increased patient engagement and vigilance, and improved patient awareness [26,27]. This topic merits further examination in future works as more data about shared notes become available.

Third, there is variation between these portals and physicians’ experiences with them, with some participants being better versed in portal usage and others having no prior experience. This, however, reflects the diverse experiences of practitioners. The generalizability of these findings is limited due to half of the studies focusing on physicians’ perceptions prior to portal use. Furthermore, while our eligibility criteria allowed for grouping of nurses, nurse practitioners, and physician assistants’ perspectives with physicians’ perspectives, this review did not address potential differences in perception. A review that captures the unique experiences of advanced practice providers with patient portals would be beneficial.

Fourth, the features offered among the portals across the studies differed. Due to the heterogeneity of portal functions, only the more common features were emphasized in this discussion. A future review that focuses on a specific feature of the inpatient portal would be helpful in capturing more nuanced opinions about inpatient portals. Lastly, most of the studies included in this review originated in the United States, which may not be directly translatable to other countries.

Overall, physicians and other health care providers acknowledge the many benefits and challenges of inpatient portals. In practice, they believed the portals improved communication and patients benefited from features such as viewing medications and scheduled appointments. However, the challenges that come with security, integration of portal use into workflow, sharing clinical notes, and allowing patient access to laboratory results pose potential barriers to portal adoption among physicians without prior portal experience. Training in use and portal-specific patient communications and expectation setting will be important to encouraging adoption among physicians. Recommendations by physicians should be considered during the design process to improve implementation and functions of the inpatient portal. In doing so, these platforms can be used more effectively to improve patient satisfaction and quality of care.