Yunan Chen, Karen Cheng, Charlotte Tang, Katie Siek, Jakob Bardram
In the era of Health 2.0, we see more novel technologies, such as mobile health applications, wearable self-tracking devices, and new communications media, being designed and developed for health consumers. With these technologies, patients can manage their own health conditions through apps, connect with other patients or health providers through online communities, and track a variety of health indicators to better understand how behavioral choices and environmental factors can influence their health. Behind such mobile, connected, and quantified-self initiatives are the notions of patient-centeredness and patient empowerment. Technologies are expected to encourage and facilitate healthy behavior change and eventually lead to positive health outcomes. Moreover, the data recorded through these sensing, tracking, and monitoring devices holds great potential for helping health providers become more aware of their patients’ conditions and make more accurate medical decisions based on that information.
In reality, however, health providers are often too busy to look at the information generated in these patient-centered systems and do not know how to deal with the novel devices their patients adopt and bring to visits. These and other related issues emerged during the panel discussion we organized at the 2013 CHI conference . The original goal of this panel was to spark discussions around the impacts of health technologies on the interaction between patients and health providers and to brainstorm ways to minimize negative impacts. However, much to our surprise, the panel triggered a lot of questions from the audience that centered on health providers’ daily practices and how we were able to study them. The discussions also triggered many comments. For example, the audience was surprised to learn that clinicians were so busy at work that they would not have time to participate in research studies or engage in the use of new patient-centered technologies. After we described how hectic clinicians’ schedules were and how little time they had for extra activities in their daily work, an audience member commented, “I always assumed that whatever we [designers/researchers] built would be a big help for them [physicians]!” This inspired us to write this article.
Why Consider Health Providers’ Needs?
The panel discussion revealed some grey areas between patient-centered practices and the expected responsibilities for health providers. Echoing the assumptions the audience made about health providers’ work, in our own research we have heard stories of patients complaining about their physicians’ reluctance and even refusal to check the health data they tracked and brought to visits. Patients also complained that their physicians did not respond to inquiries sent through communication technologies in a timely manner. Why would this happen? When we envisioned these patient-centered technologies, we believed that health providers would find it useful to know about a patient’s everyday vital signs, symptoms, exercise, diet, and other kinds of information that could influence their well-being. We assumed providers would appreciate the benefits of having this information to help them make more informed decisions. However, comments made during the panel raised issues such as that physicians do not, in general, trust “subjective” data from patients, even semi-objective data like blood pressure, as these self-recorded measures are often collected by non-certified devices and under uncontrolled circumstances. Thus, as designers, we may have misunderstood health providers’ willingness to review the data collected through such patient-centered devices.
As designers and researchers, we ask, “Is there anything missing, neglected, or invisible in the design process?” Lucy Suchman argued the importance of making work visible in information system design since “the way in which people work is not always apparent. Too often, assumptions are made as to how tasks are performed rather than unearthing the underlying work practices” . Perhaps the lack of consideration of health providers’ work practices in design, and the unrealistic expectations of their engagement in these patient-centered technologies, have led to these misconceptions. If this is the case, making health providers’ work, goals, and priorities visible for designers who are interested in these patient-driven technologies is critical.
What have We Learned?
How should we design patient-oriented systems so that health providers can use them efficiently and effectively? By efficiently, we mean they can extract relevant and important information quickly. By effectively, we mean providing the right kinds of data at the right time for them to make decisions. How do we integrate health providers’ needs in the design process to ensure that physicians’ and patients’ expectations of such systems align and the patient-centered systems provide adequate support for the tasks the health providers are expected to fulfill?
Our own experiences in studying health professionals indicated there might be inherent challenges in their work that prevented them from further engaging with these emerging patient-centered technologies. As designers, we have to consider the unique characteristics that shape health providers’ work practices, communication patterns, and constraints at work, and to design technologies that fit into their workflow. Here, we outline three main considerations based on our own experiences.
Making health providers’ work, goals, and priorities visible for designers is critical.
First, designers need to understand health providers’ work and clinical priorities and goals, and to adapt designs so the health providers can more likely benefit directly from new technologies. In one study carried out at an outpatient clinic, we spent several months shadowing health providers and talking to them during short breaks . Our observations of their work showed that it was very challenging, and often unlikely, for providers, especially physicians, to review patient-generated information in their daily practices. The main reasons are the limited availability of time and the need to perform a long list of predefined tasks. During a typical 20-minute patient visit, a physician has to complete such tasks as reviewing the patient’s previous medical history, interviewing the patient, performing a physical examination, entering orders into the electronic health record (EHR) system, and writing physician’s notes. To complete these routines in the limited time slot is already quite challenging; in more complicated patient cases, such as those who have multiple medical issues, extra time is often needed to address these problems. Hence, it was not unusual to see doctors complete their electronic charting during their lunch break and/or after their shift work.
Imagine a patient arrives with a device that contains self-tracked data over the past few months and expects the physician to examine the collected data. The patient may also invite the physician to see data from a system he has been using in monitoring his vital signs. How can these unplanned activities fit into physicians’ already packed or even backlogged schedule? There is no doubt such technologies benefit patients, but the question to consider here is what we designers and researchers can do to make these systems easy to use, not only for their intended primary users—patients—but also for health providers, who might be secondary users of these patient-centered systems. So how do we support these secondary stakeholders? Should raw data be shared with them? Should the data be displayed in the same way to providers as to patients and family members? How can patient-generated data be visualized in abstracted and meaningful ways or in customized formats to meet the needs of different providers? Understanding the unique time constraints and workflow challenges of health providers can help inform the design of patient-centered systems to minimize potential conflicts between patients who want to share large amounts of data about themselves and health providers who are often too time-pressured to look over the patients’ data.
Second, designers need to understand the different roles within the healthcare system; this can help identify providers who are likely to embrace new patient-centered technologies. It may be easy to assume that all providers are the same. Yet healthcare is a highly specialized field that encompasses many specific roles, such as physicians, nurses, medical assistants, social workers, and different therapists. Each of these roles has different medical expertise, work routines, and time constraints. Given the time constraints of physicians, they may not be the best providers to review patient-generated data and to engage in the use of patient-centered systems. Rather, it may be more beneficial for other roles, such as medical assistants and nurses, to review and summarize the collected data. If the budget permits, new positions may even be created to specifically review and handle such patient-generated data.
For instance, in a recent study, patients with bipolar disorder were asked to have their behavioral data collected by a smartphone automatically and by a questionnaire . The objective was to allow clinical staff to monitor patients and be able to react if the patients experienced signs of emerging manic or depressive episodes. But despite the fact that the system was designed with and for the clinical staff , a subsequent trial revealed that it was unclear who should be responsible for monitoring the collected health data, how often, and how to react to the data. For example, should patients’ data be monitored by a psychiatrist, a general practitioner, or a nurse, and should the data be checked weekly, daily, or every few hours? Most important, what should be done if an emergent depression or manic episode is detected? Thus, even a seemingly simple system for semi-automatic data sampling can have a significant impact on the organization of treatment and care, which should be addressed simultaneously.
Designers need to understand the liability concerns of health providers and how this may alter their willingness to engage with patient-generated data.
Similar issues were found in another study that examined an online patient portal in an outpatient clinic where a new work role was established to review patients’ messages. This new role ensured all messages were checked and responded to in a timely manner without imposing an unnecessary burden on the primary care physicians.
These two examples indicate that adopting patient-centered technologies in clinical settings is non-trivial. It is imperative for researchers to understand how new technologies fit into the existing work structure of the clinical practice, and to evaluate important issues such as the temporality of patient-generated data, levels of expertise in interpreting the data, appropriate workload, and workflows that make patient-centered technologies usable by and useful to health providers. Above all, we should not assume that providing new systems and new data would automatically benefit the work of health providers. Instead, organizational change in the health system may be required if new patient-centered technologies are to be utilized and adopted into clinical practices.
Third, designers need to understand the liability concerns of health providers and how this may alter their willingness to engage with patient-generated data. Clinical work is subject to a variety of legal regulations and is also governed by licensing boards. Stepping outside accepted guidelines or protocols may result in fines, loss of license to practice, or malpractice lawsuits. Data from patient-centered technologies is as yet unregulated by the law, and best practices have yet to be established for how to deal with such data.
In our experience, such fears can affect a health provider’s willingness to review patient-generated data. In one study, parents of preterm infants were tasked with recording their infant’s diaper usage, weight gain, and appointment attendance through a mobile app . Parents were also encouraged to track their own moods and to complete monthly assessments of their risk for postpartum depression. In the design phase of the study, we found that while the health providers liked the idea of having access to this data, they were reluctant to monitor the data on a regular basis. One concern providers raised was about overlooking important data: Would it then put the infant at risk and make the provider vulnerable to a malpractice complaint? Would the transmission of patient-generated data give parents a false sense of security, such that they are no longer as attentive to their child? And finally, the health providers were absolutely adamant that they did not want to review the emotional health data of the parents. Though it was generally agreed that parental well-being could greatly impact an infant’s health, the providers were very clear that the parents were not their patients and they did not want the legal responsibility of monitoring the parents’ data.
Our CHI panel inadvertently initiated the discussion on the design of patient-centered systems and the potential effects on health providers. In such an information-rich and evidence-driven field, designers may assume that more data will lead to better health practices. However, as we dive into the participatory Health 2.0 era, it is necessary for us to consider not only how to design such systems for patients to adopt and use for making health behavior changes, but also the secondary users whose role of reviewing, interpreting, and monitoring patient-generated data is critical to the success of these technologies. As such, the design of patient-centered systems is necessary to make the invisible work of health providers more explicit. To do so requires us to understand and be sympathetic to the work practices of health providers, and to explore ways to design the technology to benefit rather than burden them.
4. Frost, M., Doryab, A., Faurholt-Jepsen, M., Kessing, L.V., and Bardram, J.E. Supporting disease insight through data analysis: Refinements of the monarca self-assessment system. Proc. UbiComp 2013: 133–142.
6. Cheng, K.G., Hayes, G.R., Hirano, S.H., Nagel, M.S., and Baker, D. Challenges of integrating patient-centered data into clinical workflow for care of high-risk infants. Personal and Ubiquitous Computing. In press.
Yunan Chen is an associate professor in the Department of Informatics at the University of California, Irvine. Her recent projects explored the use of EHR systems in various clinical settings, with a specific focus on clinical documentation and patient-provider interaction. firstname.lastname@example.org
Karen Cheng is a senior research scientist in the Department of Informatics, University of California, Irvine, and assistant professor in the Department of Psychiatry, Charles Drew University of Medicine and Science. Her research focuses on mobile systems of health interventions and behavior change. email@example.com
Charlotte Tang is an assistant professor in computer science at the University of Michigan, Flint. Her research focuses on human-computer interaction, computer-supported cooperative work, health informatics, and accessible computing. She currently investigates the transition from a paper-based to a digital medical record system in a clinical practice. firstname.lastname@example.org
Katie A. Siek is an associate professor in the School of Informatics and Computing at Indiana University. Her primary research interests are in human-computer interaction, health informatics, and ubiquitous computing. More specifically, she is interested in how sociotechnical interventions affect personal health and well-being. email@example.com
Jakob E. Bardram is a professor in computer science at the IT University of Copenhagen, Denmark. His research focuses on human-computer interaction, computer-supported cooperative work, and ubiquitous computing, with a special focus on applications within healthcare. He is currently researching how patient-focused mobile technology can help mentally ill patients. firstname.lastname@example.org
©2014 ACM 1072-5220/14/09 $15.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2014 ACM, Inc.