Designing Inpatient Portals to Support Patient Agency and Dynamic Hospital Experiences

Abstract

Inpatient portals could help patients engage in their hospital care, yet several design, usability, and adoption issues prevent this technology from fulfilling its potential. Despite patients having needs that extend beyond the scope of existing inpatient portals, we know less about how to design such portals that support them. To learn about effective designs, we created three mid-fidelity prototypes representing novel approaches for inpatient portal design. Then, we conducted interviews with 21 pediatric and adult inpatients to gather their feedback on these prototypes. Participants shared how the prototypes addressed the following needs: forming active partnerships, navigating relationships and power dynamics with clinicians, understanding complexity of care, contextualizing health information, increasing efficiency of communication, and preventing lost information. We discuss two key implications—supporting patients’ agency and dynamic needs throughout their hospital care—for the future of inpatient portal designs.

Introduction

The hospital is universally recognized as one of the most disempowering settings in the United States healthcare system.⁽¹⁾ In addition to the stress of being ill, patients and caregivers must communicate with large care teams, handle social isolation, be watchful for medical errors, and keep up with ever-changing health information.^(2–4) To help overcome such challenges and increase patient engagement, health informatics researchers have investigated the design and use of inpatient-facing technologies—including information displays, medication reconciliation tools, and digital checklists—for the hospital setting.^(5–8)

One type of patient-facing technology is the inpatient portal, which gives patients and caregivers access to their medication history, test results, clinical notes, educational content, and secure messaging with providers. Such features have helped patients take on a more active role in discussions with providers, make informed care decisions, and speak up to correct mistakes in their Electronic Health Record (EHR).^(9–11) Despite known benefits, adoption rates of inpatient portals remain low.^(12,13) The level and timeliness of information shared with patients via the portal is often restricted,⁽¹⁴⁾ and patients have identified several design, acceptance, and usability issues.^(15–17) Researchers have also recognized disparities in portal use.^(18,19) As a result, inpatient portals have shown limited impact on patient engagement.^(20–22)

Because inpatient portals were adapted from EHR systems that were originally designed for providers as end-users, these technologies do not fully support patients’ information and experiential needs. In a prior low-fidelity prototype study, we found that patients had needs and values—such as flagging problems and preparing for at-home care—that extend far beyond the utility of current inpatient portals.⁽²³⁾ Yet questions still remain about how to address these needs and values in cohesive inpatient portal designs, and how patients envision using such designs during their hospital stay. To answer these questions, we created 3 mid-fidelity prototypes of novel inpatient portal designs and gathered feedback from 21 inpatients at an adult and a pediatric hospital. In this paper, we share additional inpatient needs that the mid-fidelity prototypes uncovered, our participants’ perspectives on how the mid-fidelity prototypes aligned with their hospital experiences, and how participants envisioned using the prototypes during their hospital stay. We then discuss key implications and future directions for inpatient portal design.

Methods

Our goal was to investigate how inpatient portals could be designed to support the needs of inpatients during their hospital stay. We sought feedback from patients on mid-fidelity prototypes depicting inpatient portal designs.

Mid-Fidelity Prototypes

To further explore patients’ inpatient portal needs, we conducted an iterative design process using Sketch¹ and Marvel² to create three different prototypes (Timeline, Goal, and Categorical) for a tablet-based inpatient portal. Each prototype emphasized one aspect of patients’ values and needs based on the design guidelines from our previous low-fidelity prototype study.⁽²³⁾ Based on the findings from this previous study, we also included a core set of features—contacting caregivers, messaging providers, and setting reminders—in all these prototypes.

The Timeline prototype (Figure 1) emphasizes the need for patients to view timing and details of anticipated care events. It presents the patient’s schedule for the day in the form of a timeline that allows people to click on any event and obtain more information about this event. For example, selecting the “Vitals Check” event shows a description of what will happen during this event, as well as the patient’s most recent vital measures (e.g., temperature, heart rate, and blood pressure). Additionally, the name of the provider who is involved in the event is shared, as well as their documented notes. The last event listed on the patient’s schedule is their projected discharge date.

Figure 1.

Screenshot of the Timeline prototype.

The Goal prototype (Figure 2) highlights the patient’s and provider’s goals along with the tasks that the patient must complete to meet each goal. For example, clicking on the “Recover from Surgery” goal reveals the details of the necessary tasks and care-related information (e.g., relevant medications, providers responsible for helping the patient complete the task). Above the goals and tasks section sits two dynamic sliding scales, each representing the patient’s and provider’s assessment of the patient’s readiness to be discharged from the hospital. Within the prototype, the patient can change their sliding scale to communicate their perspective to the care team.

Figure 2.

Screenshot of the Goal prototype.

Finally, the Categorical prototype (Figure 3) highlights categories of care information—such as medications, test results, and discharge details—and is most similar to current portal designs. Clicking on the “Dashboard” shows the care team’s activity (e.g., when a provider shift change occurs) and most recent modifications to their care plan (e.g., when an order for a new medication is placed).

Figure 3.

Screenshot of the Categorical prototype.

Study Setting, Eligibility, Recruitment, and Procedures

The study took place at two hospitals in a metropolitan area of the United States: a pediatric and an adult hospital site. Each hospital manages 300-400 beds and admits approximately 15,000 inpatients annually. All study procedures were IRB approved. Patients were considered eligible for this study if they were at least 7 years old (an age when children can self-reflect, reason, conceptualize, and articulate their thoughts about technologies),⁽²⁴⁾ could communicate in English, and were well enough to provide informed consent. To seek a breadth of perspectives, we approached eligible inpatients across a variety of medical and surgical services. We also obtained consent from parents and legal guardians (i.e., caregivers) of the pediatric patients who assented to enroll in the study. While our interviews prioritized the opinion of patients, caregivers who were physically present during these interviews were also invited to share their thoughts.

We conducted semi-structured interviews with participants during their hospital stay to gather their input on the information, features, and design of the three prototypes. Two research team members were present during interviews to ask questions and take observational notes. We showed participants the three prototypes in random order on an iPad. After a brief explanation of the prototypes, we asked participants to think aloud while they explored and compared the prototypes’ features. We then asked the participant to describe the suitability of using the prototypes during their hospital stay. Interviews typically lasted 40 minutes, were audio recorded, and transcribed for analysis. All participants were compensated $25 for their time.

Data Analysis

With a subset of transcripts, multiple research team members did several rounds of independent open-coding, meeting between each round to discuss emergent themes and develop a preliminary inductive codebook.⁽²⁵⁾ One research team member then took this codebook and applied it to the remainder of the transcripts, adding or revising themes as appropriate, and discussing the process with the research team. Through these discussions, the research team identified relevant themes from prior work regarding inpatient engagement (e.g., attitudes towards partnership, relationships with clinicians) that further scaffolded our analysis.⁽²⁶⁾ Two research team members used these themes as axial codes and iteratively coded, discussed interpretations, and revised code definitions until consensus was reached.

Participants

We recruited at both study sites until the interviews stopped yielding new information (i.e., saturation was reached), enrolling a total of 21 participants. Eleven participants were pediatric patients (P1-11) between 7-17 years old. The remaining 10 participants were adult patients (A1-A10) between 18-79 years old. Eleven participants identified as female and 10 participants identified as male. Fifteen participants identified as White/Caucasian, 2 as Black/African American, 2 as Hispanic/Latinx, 1 as Native Hawaiian/Pacific Islander, and 1 participant reported their ethnicity as “other”. Participants had diverse clinical backgrounds supported across medical and surgical hospital services. Five of the 21 participants (24%) were first-time hospital patients at the time of their interview.

Results

We used our interviews to gather participant feedback on the design and features of the mid-fidelity prototypes. Overall, 9/21 participants (43%) preferred the Categorical prototype, compared to the Timeline (4/21, or 19% participants) and Goal (4/21, or 19% participants) prototypes. The remaining 4/21 participants did not prefer a particular prototype. Despite stronger favor for the Categorical prototype, our participants discussed how aspects of all three prototypes addressed the following needs: forming active partnerships, navigating relationships and power dynamics with clinicians, understanding complexity of care, contextualizing health information, increasing efficiency of communication, and preventing lost information.

Forming Active Partnerships

Our participants discussed several ways in which the mid-fidelity prototype features could help them have a more active partnership with their care team. A9 mentioned being the point-person to coordinate handoffs between the doctor and nurse when the care team didn’t have the most up-to-date information. A9 shared how the Categorical prototype’s team activity feature could help fulfill their point-person role in this scenario: “Like the doctor comes in this morning and he gives me a brief on what's going on, and then the nurse comes in and I have more information than she does...if [the doctor] puts it in the chart somewhere on the computer and he didn't tell me, so now the nurse is wanting something different from what I expect, it'd be nice if we have all that information available there.”

Aside from coordinating handoffs between providers, participants spoke about how certain features could show what patients could do to help themselves during and after their hospital stay. P6 chose the Timeline prototype as helping her to monitor upcoming blood draws, anticipate their results, and prepare for next steps in her care: “I would make a timeline, do blood draws every four hours...until they work, because you need to get the results. I think it's really important for people to know what they need to do and what they can do.”

Setting personalized goals in the Goal prototype was often cited by participants as a way to take charge, be proactive with their health care, and collaborate with their care team. While viewing the Goal prototype, A8 discussed how she appreciated the feature to add and set her own goals. She explained, “[it’s] irritating when someone says, ‘what are your goals?’ And they're writing them down and then they give them to you, and you fold up the paper and throw it in your bag to leave...And then you get home - you're like, I should have been doing these goals, but oh well, throw it in the garbage, back to my regular life. This app wouldn't allow you to do that, this app would keep you interested.”

Some pediatric patients in our study also liked the ability to set goals to independently manage their care. P4 saw their care team as partnering with patients to decide what health goals should be set: “Some times are just like once you finish this medicine, you can leave...Other times, take this much of something every day...So this is nice that there's something to take care of yourself at home, but I would suggest somehow having the doctors kind of help start the list, because it can be like someone sets themselves too low or too high, the doctors can really help. They know you and they can help too.”

In contrast, A7 did not see setting goals as being particularly useful, stating: “…it’s interesting, but it’s only one small part of the whole hospital stay.” A few pediatric participants thought this feature could reduce their active partnerships with clinicians. P7 and P2 thought that setting goals—specifically discharge goals—could risk setting false care expectations and cause disappointment when these expectations are not met. From their perspective, the uncertain nature of the hospital stay meant that a patient’s ability to claim ownership over their health was sometimes in flux and unpredictable.

Navigating Relationships and Power Dynamics with Clinicians

Participants spoke about how the mid-fidelity prototypes could help establish a human-based—rather than clinical-based—relationship with their providers by exerting authority over personal aspects of their stay. P9 discussed how he could use the Timeline prototype’s schedule of the day to have more control over his day-to-day life in the hospital: “right now all I’m seeing here is straight hospital labs and medical stuff. And that’s not bad, but as a patient, you’re more than just a guinea pig, you’re a human still...So I think like if I would add something, it would be like I’m scheduling a break here, I want to see my visits here, or someone wants to come visit me...you want it to be a fully functional app to the point where it’s [used] for all things, not just your medical.”

For P8’s caregiver, building personal connections with providers was one way to humanize the hospital experience and not be treated as a passive recipient of care. This father wanted to add a feature in the Goal prototype so he could “provide feedback” to the care team and establish a personal relationship with them. P8’s caregiver explained, “if there’s a way to - ‘hey, I like to play tennis on the weekends.’ Like who are they? That would be kind of cool, because it may make the patients feel a little better because I have a personal connection with people, you know?”

Other participants saw the mid-fidelity prototypes as helping them to question or challenge their clinician’s judgement without causing personal conflicts. A8 described a past experience where she was very “upset” that her providers missed examining a spot on her kidney that later developed into renal cancer. She remembered thinking during this incident: “Doctor, you're the expert, but this is my body and I feel like if there's a spot there that's not supposed to be there, you need to check on that.” Based on this negative experience, A8 expressed her appreciation for the Timeline prototype’s report concern feature to have this conversation with providers in a non-confrontational manner: “I would want to report that [concern] to my doctor - I didn't get a clear answer, and not to bug you, but could you please explain again.”

A1 had a similar view about the Goal prototype’s assessment scales, sharing how this feature could mitigate asymmetrical power dynamics by surfacing conflicting opinions between patients and providers: “ultimately it would be up to your doctor to decide [if you’re ready for discharge], but it’s so neat that as a patient I could have a voice...because a lot of times I feel like I don’t get a voice...I like that you can read what the team thinks and why they think you’re ready or not ready, and then you can compare and kind of go from there.” For both A8 and A1, the prototypes allowed them to navigate the line between engaging in their hospital care and deferring to their providers’ expertise. However, A10 did not want features like the team assessment scale to detract from his already strong relationship with his care team: “I would be interested in their assessment, and I would take it into careful consideration…what I have in terms of the relationship, I don’t think I would [use this feature]. For me the personal face-to-face interaction…works best for me.”

Understanding Complexity of Care

Our participants discussed how detailed explanations in the prototypes could help them overcome their struggles to understand complex clinical details and reduce confusion during their hospital stay. Beyond raw test results, patients wanted some way to understand why certain care decisions were considered or made by providers, and to interpret what the impact of these decisions might be. While clicking through the prototypes, A5 explained: “What I was thinking was, do you have an explanation of what the test is and you’ve got one right here...so that’s good. Because they take so many blood tests and I don’t have a frigging clue about what a lot of it is and what it pertains to and what the impact is.” On the other hand, P11 thought these explanations were redundant to simply Googling the information, a practice he already did to understand the purpose and side effects of his medications.

Participants identified their care team’s notes as an important resource within the prototypes for understanding their care. However, opinions varied on the level of detail they wanted from the notes. Some participants wanted access to all content in their clinical notes, but others wanted a digestible summary of key information and next steps. A2 said: “I want to see the main key points, like the gist of what's going on, just the important details. For instance, we're going to take her off her pain medicine drip because her heart rate was slowing down, like the juicy tidbits.” P10 only wanted to view text-based notes, as supplementary MRI and X-Ray images of inside the body were thought to be disturbing. P4 suggested changing the design of the prototypes to include an indicator or note from providers to prevent unnecessary worry about test results: “I would want a doctor to say ‘everything's okay for now, I will update you later’ type thing if there's nothing to say, and then if something's urgent, say it's urgent, because if it's saying more testing, you start thinking about what they're testing. I'm a pretty anxious person with that type of stuff, but I think most people are with medicine.”

Contextualizing Health Information

Aside from understanding the clinical language their providers use to communicate, our participants expressed their need for additional resources to contextualize information about their health, beyond their discussions with providers. For example, participants wanted to use their own medical history and information from other patients to interpret their progress based on what was considered “normal”. A few participants raised this idea in reference to the patient and provider assessment scales within the Goal prototype. A5 said: “it’s a good way to track your progress…[test result] varies so much and sometimes your benchmark isn’t what somebody else’s would be.” A8 wanted to change the Goal prototype’s medication list to include similar benchmark information: “You know what else I would like to see...is what would be like the normal? Because...when I get my blood pressure taken - well, what's the normal rate compared to what I'm doing? What's the normal temperature that I'm supposed to have? What's the norm for me compared to the normal?” When asked what she meant by “the normal,” A8 clarified, “the normal range for me and then the normal range against other healthy citizens.”

Participants also discussed their need to access information that they currently could only obtain with the help of their providers. For example, P2 shared his trial-and-error experience in learning what he could order from the hospital’s cafeteria menu, given his dietary restrictions during treatment. When asked about what changes he would make to improve all three prototypes, he said he wanted an efficient way to identify what hospital food options were compatible with his treatment and send his food order to the kitchen. P2 explained further, “usually I get a pamphlet, and then I just have to go through everything and check it off...it would be best if the food that you can’t eat just doesn’t pop up on there...you can order it, but I asked my nurse, and she says they’ll just deny it.” From P2’s point of view, surfacing this hidden information could have prevented his frustrating experience.

Increasing Efficiency of Communication

Because verbal communication remains common in the hospital, participants discussed ways for the prototypes to help optimize communication between patients, caregivers, and providers. While exploring our mid-fidelity prototypes, participants thought direct messages with providers—coupled with estimated response times—could help get answers to their questions without having to wait for providers to physically be in the hospital room. P7 discussed how the prototypes’ messaging feature would allow providers to prepare answers to questions before visiting the patient’s hospital room: “Sometimes you want to talk to your nurse, but she’s stuck in another room, so she can’t really get to you right away…[messaging] would be more useful, because then before they come in...they could bring material...it would just take less time too.” Although A10 did not think this messaging feature was essential to his hospital stay, he thought it could help with timing his conversations: “I don’t want to put a message out there and just let it be hanging out in the ether somewhere, and not hear anything back. So I think that would be an enhancement to what I’m currently doing.”

In addition, many pediatric participants thought the Timeline prototype’s schedule of the day could help coordinate the timing of visits and physical presence of their caregivers. For example, P4 thought the feature might help their caregivers know when it was alright to physically leave the room without missing critical information: “If the parent wants to go out and get food or coffee but they want to hear what’s going on with their kid, it’s really good for them to know.” Other pediatric patients thought that using this schedule to coordinate with their parents was a way to ensure they had a supportive presence in the room when they received care information from providers. As P2 described, “sometimes you get a result back and [family’s] not there and you’re like - oh, how do I tell them that they should be home or I might be getting surgery and you’re like oh, that would be nice if [family] were here.”

Preventing Lost Information

Several participants identified aspects of the mid-fidelity prototypes that could support them in preventing information that is lost when their expertise is ignored by providers, their concerns are not communicated between shift changes, or their input is not sought before a care decision. When viewing the medication list within the Goal prototype, P1 wanted a feature to confirm that the care team was aware of her medication allergies. A1 explained how she would use the direct message feature within the Goal prototype to ensure her more urgent concerns were relayed to providers: “like what if your goals didn’t match the doctor’s goals...unfortunately I’m one of those people that always needs more [pain medication] than the average person. It’s frustrating, because some doctors are like ‘this should work’ and I’m telling you it’s not, my blood pressure is telling you it’s not - please help me, I’m in pain...so it would be nice to have a spot where you could be like ‘hey...am I going to get any help with this...can I speak to someone?’ That sort of thing.”

Other features were thought to help patients identify missing information that could cause mistakes in their care. For this reason, A5 discussed the value of patients having access to clinical notes within the prototypes: “I haven’t been on the [patient] portal since I’ve been in here, but they change medications and [providers] don’t necessarily get the information...they bring me my pills last night, I normally take Warfarin - there’s no Warfarin in it. And they go, ‘it’s not on the list here.’ I said, ‘well you need to call somebody and get it on the list because that’s why I’m here.’...it’s just a matter of the doctor probably forgot or whatever. So if you could go look [at the notes], it would help.”

Indeed, participants believed features that documented care decisions would allow them to report issues, confirm their requests were dealt with, and cross-check or correct the information that providers shared with them. P11’s caregiver described a weekend in the hospital where a different provider was on call to handle their case and made a decision that deviated from the team’s agreed-upon plan. The change “devastated” and “traumatized” P11. About this incident, P11’s caregiver said: “I think [access to our information] probably would have been helpful, because then we could say, ‘no, look, this is what we’re doing, this is the plan, we’re not going to change it.’ It’s our word against - you know, there’s nothing set in stone of okay, this is what the doctor said...if we had this, like the Timeline [prototype], it would show what the plan is for what we’re doing, so that everybody is all on the same page.”

Discussion

Our interviews allowed participants to share their perspectives on how future inpatient portals could meet their needs and overcome challenges to engaging in their hospital care. Seeking open-ended feedback from participants on the mid-fidelity prototypes revealed additional needs and values that could not otherwise be captured in later stages of the inpatient portal design process (i.e., usability studies, post-implementation questionnaires).⁽²⁷⁾ Below, we share two key design implications for future inpatient portals.

First, future inpatient portals should be designed to support patients’ agency during their hospital stay. Most inpatient portals effectively deliver information to patients, but do not always offer a space for patients to contribute their own expertise or communicate their care needs.⁽²⁸⁾ Although one participant already felt this agency through face-to-face interactions with his care team, many other participants valued features—including setting goals, reporting concerns, and adding scheduled events—that allowed them to provide input and that gave them a greater sense of control over their lived experience in the hospital. Moreover, participants saw features as helping to overcome barriers to their agency. A1 saw the Goal prototype’s assessment scales as a way to navigate unequal power dynamics with clinicians and discuss her concerns; P11’s caregiver thought a documented care plan in the Timeline prototype would have prevented lost information and convinced clinicians to follow the agreed-upon plan for her child. These findings suggest an opportunity to hybridize these features into a cohesive system that supports patients’ agency during their hospital stay. Future inpatient portals could offer patients options to edit their health goals and daily schedules, to prioritize concerns they want addressed, to share their own assessment of their treatment progress, and to access a documented history of care team decisions. These features would help inpatient portals become tools for empowerment, extending its current purpose beyond simply delivering clinical information to the patient.

Second, future inpatient portals should accommodate patients’ changing needs and engagement. Our interviews with participants revealed the dynamic, rather than static, nature of their hospital stay. Two participants thought goal-setting features in future inpatient portals could allow them to have an active partnership with their care team, but their uncertain health conditions might impact progress toward their goals during their stay. Participants also varied in how much health information they preferred to access and act upon while in the hospital. These findings support prior work’s assertion that a patient’s engagement in their healthcare can continuously evolve and shift.⁽²⁹⁾ Inpatient portals must therefore reflect patients’ dynamic needs and preferences. Researchers have previously leveraged adaptive and customizable views of tablet-based cancer navigation tools.⁽³⁰⁾ Similarly, future inpatient portals could prompt patients to specify what content they want to view (e.g., medications, schedule of the day, personal health goals), display this content, and allow patients to change their content preferences as needed. Updates to the content (e.g., revising a health goal, adjustment to a medication dose) could be highlighted within the portal as patients progress through their treatment plan. This dynamic approach would not only help patients focus on the health information that matters to them but would also accommodate the patient’s continuously evolving hospital experience.

Lastly, we recognize the limitations of our study. We sought input from inpatient participants at two hospital sites within one metropolitan area of the United States. Therefore, our findings might not be transferrable to other patient populations or geographic regions. Because study participation was voluntary, views of patients who chose to participate might not reflect those who were not approached or decided against participation. Study team members who were involved in creating the prototypes also participated in interviews with patients, which may have elicited more positive feedback than neutral interviewers.⁽³¹⁾ The novelty effect might have also influenced participant responses to the prototypes.

Aside from these limitations, our study presents many opportunities for future work. Given the increasing standardization of commercial inpatient portals,⁽³²⁾ additional research is needed to explore potential differences in needs and designs for subgroups of hospital services, diagnoses, treatments, and pediatric vs. adult patients. Testing high-fidelity versions of our prototypes with patients’ actual health data would also help researchers evaluate their real-world impact on hospital stays. Further input from clinical stakeholders on how to address patient needs while also ensuring future portals fit within the existing technical infrastructure of hospitals is another important future step.

Conclusion

In this paper, we present a mid-fidelity prototype study to improve the design of inpatient portals and meet the needs of patients. Pediatric and adult participants discussed our prototypes as meeting the following needs: forming active partnerships, navigating relationships and power dynamics with clinicians, understanding complexity of care, contextualizing health information, increasing efficiency of communication, and preventing lost information. Our findings show the importance of designing inpatient portals to support patients’ agency, as well as their dynamic needs and preferences. By making such considerations in future inpatient portals, these technologies can begin to fulfill their potential of helping patients increase their engagement and improve their hospital experiences.

Acknowledgements

We thank our participants and study coordinators for their contributions. This work was primarily funded by AHRQ #1R01HS022894 and in part by NIMH #T32MH115882. All authors were involved in study conception, analysis, and paper writing/edits; MK in designing the study and prototypes; SH, MK, SRM, and CA in data collection.